Using GluonCV 0.3 with Apache MXNet 1.3 source code: https://github.com/tspannhw/nifi-gluoncv-yolo3 *Captured and Processed Image Available for Viewing in Stream in Apache NiFi 1.7.x use case: I need to easily monitor the contents of my security vault. It is a fixed number of known things. What we need in the real world is a nice camera(s) (maybe four to eight depending on angles of the room), a device like an NVidia Jetson TX2, MiniFi 0.5 Java Agent, JDK 8, Apache MXNet, GluonCV, Lots of Python Libraries, a network connection and a simple workflow. Outside of my vault, I will need a server(s) or clusters to do the more advanced processing, though I could run it all on the local box. If the number of items or certain items I am watching are no longer in the screen, then we should send an immediate alert. That could be to an SMS, Email, Slack, Alert System or other means. We had most of that implemented below. If anyone wants to do the complete use case I can assist. demo implementation: I wanted to use the new YOLO 3 model which is part of the new 0.3 stream, so I installed a 0.3. This may be final by the time you read this. You can try to do a regular pip3.6 install -U gluoncv and see what you get. pip3.6 install -U gluoncv==0.3.0b20180924 Yolo v3 is a great pretrained model to use for object detection. See: https://gluon-cv.mxnet.io/build/examples_detection/demo_yolo.html The GluonCV Model Zoo is very rich and incredibly easy to use. So we just grab the model "yolo3_darknet53_voc" with an automatic one time download and we are ready to go. They provide easy to customize code to start with. I write my processed image and JSON results out for ingest by Apache NiFi. You will notice this is similar to what we did for the Open Computer Vision talks: https://community.hortonworks.com/articles/198939/using-apache-mxnet-gluoncv-with-apache-nifi-for-de.html This is updated and even easier. I dropped the MQTT and just output image files and some JSON to read. GluonCV makes working with Computer Vision extremely clean and easy. why Apache NiFi For Deep Learning Workflows Let me count the top five ways: #1 Provenance - lets me see everything, everywhere, all the time with the data and the metadata. #2 Configurable Queues - queues are everywhere and they are extremely configurable on size and priority. There's always backpressure and safety between every step. Sinks, Sources and steps can be offline as things happen in the real-world internet. Offline, online, wherever, I can recover and have full visibility into my flows as they spread between devices, servers, networks, clouds and nation-states. #3 Security - secure at every level from SSL and data encryption. Integration with leading edge tools including Apache Knox, Apache Ranger and Apache Atlas. See: https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.1.1/bk_security/content/ch_enabling-knox-for-nifi.html #4 UI - a simple UI to develop, monitor and manage incredibly complex flows including IoT, Deep Learning, Logs and every data source you can throw at it. #5 Agents - MiniFi gives me two different agents for my devices or systems to stream data headless. running gluoncv yolo3 model I wrap my Python script in a shell script to throw away warnings and junk cd /Volumes/TSPANN/2018/talks/ApacheDeepLearning101/nifi-gluoncv-yolo3 python3.6 -W ignore /Volumes/TSPANN/2018/talks/ApacheDeepLearning101/nifi-gluoncv-yolo3/yolonifi.py 2>/dev/null List of Possible Objects We Can Detect ["aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow", "diningtable", "dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"] I am going to train this with my own data for the upcoming INTERNET OF BEER, for the vault use case we would need your vault content pictures. See: https://gluon-cv.mxnet.io/build/examples_datasets/detection_custom.html#sphx-glr-build-examples-datasets-detection-custom-py Example Output in JSON {"imgname": "images/gluoncv_image_20180924190411_b90c6ba4-bbc7-4bbf-9f8f-ee5a6a859602.jpg", "imgnamep": "images/gluoncv_image_p_20180924190411_b90c6ba4-bbc7-4bbf-9f8f-ee5a6a859602.jpg", "class1": "tvmonitor", "pct1": "49.070724999999996", "host": "HW13125.local", "shape": "(1, 3, 512, 896)", "end": "1537815855.105193", "te": "4.199203014373779", "battery": 100, "systemtime": "09/24/2018 15:04:15", "cpu": 33.7, "diskusage": "49939.2 MB", "memory": 60.1, "id": "20180924190411_b90c6ba4-bbc7-4bbf-9f8f-ee5a6a859602"} Example Processed Image Output It found one generic person, we could train against a known set of humans that are allowed in an area or are known users. nifi flows Gateway Server (We could skip this, but aggregating multiple camera agents is useful) Send the Flow to the Cloud Cloud Server Site-to-Site After we infer the schema of the data once, we don't need it again. We could derive the schema manually or from another tool, but this is easy. Once you are done, then you can delete the InferAvroSchema processor from your flow. I left mine in for your uses if you wish to start from this flow that is attached at the end of the article. flow steps Route When No Error to Merge Record Then Convert Those Aggregated Apache Avro Records into One Apache ORC file. Then store it in an HDFS directory. Once complete their will be a DDL added to metadata that you can send to a PutHiveQL or manually create the table in Beeline or Zeppelin or Hortonworks Data Analytics Studio (https://hortonworks.com/products/dataplane/data-analytics-studio/). schema: gluoncvyolo { "type" : "record", "name" : "gluoncvyolo", "fields" : [ { "name" : "imgname", "type" : "string", "doc" : "Type inferred from '\"images/gluoncv_image_20180924211055_8f3b9dac-5645-49aa-94e7-ee5176c3f55c.jpg\"'" }, { "name" : "imgnamep", "type" : "string", "doc" : "Type inferred from '\"images/gluoncv_image_p_20180924211055_8f3b9dac-5645-49aa-94e7-ee5176c3f55c.jpg\"'" }, { "name" : "class1", "type" : "string", "doc" : "Type inferred from '\"tvmonitor\"'" }, { "name" : "pct1", "type" : "string", "doc" : "Type inferred from '\"95.71207000000001\"'" }, { "name" : "host", "type" : "string", "doc" : "Type inferred from '\"HW13125.local\"'" }, { "name" : "shape", "type" : "string", "doc" : "Type inferred from '\"(1, 3, 512, 896)\"'" }, { "name" : "end", "type" : "string", "doc" : "Type inferred from '\"1537823458.559896\"'" }, { "name" : "te", "type" : "string", "doc" : "Type inferred from '\"3.580893039703369\"'" }, { "name" : "battery", "type" : "int", "doc" : "Type inferred from '100'" }, { "name" : "systemtime", "type" : "string", "doc" : "Type inferred from '\"09/24/2018 17:10:58\"'" }, { "name" : "cpu", "type" : "double", "doc" : "Type inferred from '12.0'" }, { "name" : "diskusage", "type" : "string", "doc" : "Type inferred from '\"48082.7 MB\"'" }, { "name" : "memory", "type" : "double", "doc" : "Type inferred from '70.6'" }, { "name" : "id", "type" : "string", "doc" : "Type inferred from '\"20180924211055_8f3b9dac-5645-49aa-94e7-ee5176c3f55c\"'" } ] } Tabular data has fields with types and properties. Let's specify those for automated analysis, conversion and live stream SQL. hive table schema: gluoncvyolo CREATE EXTERNAL TABLE IF NOT EXISTS gluoncvyolo (imgname STRING, imgnamep STRING, class1 STRING, pct1 STRING, host STRING, shape STRING, end STRING, te STRING, battery INT, systemtime STRING, cpu DOUBLE, diskusage STRING, memory DOUBLE, id STRING) STORED AS ORC; Apache NiFi generates tables for me in Apache Hive 3.x as Apache ORC files for fast performance. hive acid table schema: gluoncvyoloacid CREATE TABLE gluoncvyoloacid (imgname STRING, imgnamep STRING, class1 STRING, pct1 STRING, host STRING, shape STRING, `end` STRING, te STRING, battery INT, systemtime STRING, cpu DOUBLE, diskusage STRING, memory DOUBLE, id STRING) STORED AS ORC TBLPROPERTIES ('transactional'='true') I can just as easily insert or update data into Hive 3.x ACID 2 tables. We have data, now query it. Easy, no install analytics with tables, Leafletjs, AngularJS, graphs, maps and charts. nifi flow registry To manage version control I am using the NiFi Registry which is great. In the newest version, 0.2, there is the ability to back it up with github! It's easy. Everything you need to know is in the doc and Bryan Bend's excellent post on the subject. https://nifi.apache.org/docs/nifi-registry-docs/index.html https://bryanbende.com/development/2018/06/20/apache-nifi-registry-0-2-0 There were a few gotchas for me. Use your own new github project with permissions and then clone it local git clone https://github.com/tspannhw/nifi-registry-github.git Make sure github directory has permission and is empty (no readme or junk) Make sure you put in the full directory path Update your config like below: <flowPersistenceProvider> <class>org.apache.nifi.registry.provider.flow.git.GitFlowPersistenceProvider</class> <property name="Flow Storage Directory">/Users/tspann/Documents/nifi-registry-0.2.0/conf/nifi-registry-github</property> <property name="Remote To Push">origin</property> <property name="Remote Access User">tspannhw</property> <property name="Remote Access Password">generatethis</property> </flowPersistenceProvider> This is my github directory to hold versions: https://github.com/tspannhw/nifi-registry-github resources: https://github.com/tspannhw/UsingGluonCV https://gluon.mxnet.io/chapter01_crashcourse/ndarray.html https://gluon-cv.mxnet.io/build/examples_detection/demo_yolo.html#sphx-glr-build-examples-detection-demo-yolo-py https://gluon-cv.mxnet.io/model_zoo/index.html#object-detection https://community.hortonworks.com/articles/215271/iot-edge-processing-with-deep-learning-on-hdf-32-a-2.html https://community.hortonworks.com/articles/198912/ingesting-apache-mxnet-gluon-deep-learning-results.html zeppelin notebook apache-mxnet-gluoncv-yolov3-copy.json nifi flow gluoncv-server.xml ... View more

Running Apache MXNet Deep Learning on YARN 3.1 - HDP 3.0 With Hadoop 3.1 / HDP 3.0, we can easily run distributed classification, training and other deep learning jobs. I am using Apache MXNet with Python. You can also do TensorFlow or Pytorch. If you need GPU resources, you can specify them as such: yarn.io/gpu=2 My cluster does not have an NVidia GPU unfortunately. See: https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/data-operating-system/content/dosg_recommendations_for_running_docker_containers_on_yarn.html Running App on YARN [root@princeton0 ApacheDeepLearning101]# ./yarn.sh 18/09/21 15:31:22 INFO distributedshell.Client: Initializing Client 18/09/21 15:31:22 INFO distributedshell.Client: Running Client 18/09/21 15:31:22 INFO client.RMProxy: Connecting to ResourceManager at princeton0.field.hortonworks.com/172.26.208.140:8050 18/09/21 15:31:23 INFO client.AHSProxy: Connecting to Application History server at princeton0.field.hortonworks.com/172.26.208.140:10200 18/09/21 15:31:23 INFO distributedshell.Client: Got Cluster metric info from ASM, numNodeManagers=1 18/09/21 15:31:23 INFO distributedshell.Client: Got Cluster node info from ASM 18/09/21 15:31:23 INFO distributedshell.Client: Got node report from ASM for, nodeId=princeton0.field.hortonworks.com:45454, nodeAddress=princeton0.field.hortonworks.com:8042, nodeRackName=/default-rack, nodeNumContainers=4 18/09/21 15:31:23 INFO distributedshell.Client: Queue info, queueName=default, queueCurrentCapacity=0.4, queueMaxCapacity=1.0, queueApplicationCount=8, queueChildQueueCount=0 18/09/21 15:31:23 INFO distributedshell.Client: User ACL Info for Queue, queueName=root, userAcl=SUBMIT_APPLICATIONS 18/09/21 15:31:23 INFO distributedshell.Client: User ACL Info for Queue, queueName=root, userAcl=ADMINISTER_QUEUE 18/09/21 15:31:23 INFO distributedshell.Client: User ACL Info for Queue, queueName=default, userAcl=SUBMIT_APPLICATIONS 18/09/21 15:31:23 INFO distributedshell.Client: User ACL Info for Queue, queueName=default, userAcl=ADMINISTER_QUEUE 18/09/21 15:31:23 INFO distributedshell.Client: Max mem capability of resources in this cluster 15360 18/09/21 15:31:23 INFO distributedshell.Client: Max virtual cores capability of resources in this cluster 12 18/09/21 15:31:23 WARN distributedshell.Client: AM Memory not specified, use 100 mb as AM memory 18/09/21 15:31:23 WARN distributedshell.Client: AM vcore not specified, use 1 mb as AM vcores 18/09/21 15:31:23 WARN distributedshell.Client: AM Resource capability=<memory:100, vCores:1> 18/09/21 15:31:23 INFO distributedshell.Client: Copy App Master jar from local filesystem and add to local environment 18/09/21 15:31:24 INFO distributedshell.Client: Set the environment for the application master 18/09/21 15:31:24 INFO distributedshell.Client: Setting up app master command 18/09/21 15:31:24 INFO distributedshell.Client: Completed setting up app master command {{JAVA_HOME}}/bin/java -Xmx100m org.apache.hadoop.yarn.applications.distributedshell.ApplicationMaster --container_type GUARANTEED --container_memory 512 --container_vcores 1 --num_containers 1 --priority 0 1><LOG_DIR>/AppMaster.stdout 2><LOG_DIR>/AppMaster.stderr 18/09/21 15:31:24 INFO distributedshell.Client: Submitting application to ASM 18/09/21 15:31:24 INFO impl.YarnClientImpl: Submitted application application_1536697796040_0022 18/09/21 15:31:25 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=AM container is launched, waiting for AM container to Register with RM, appMasterHost=N/A, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=ACCEPTED, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:26 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=AM container is launched, waiting for AM container to Register with RM, appMasterHost=N/A, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=ACCEPTED, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:27 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=AM container is launched, waiting for AM container to Register with RM, appMasterHost=N/A, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=ACCEPTED, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:28 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=RUNNING, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:29 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=RUNNING, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:30 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=RUNNING, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:31 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=RUNNING, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:32 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=RUNNING, distributedFinalState=UNDEFINED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:33 INFO distributedshell.Client: Got application report from ASM for, appId=22, clientToAMToken=null, appDiagnostics=, appMasterHost=princeton0/172.26.208.140, appQueue=default, appMasterRpcPort=-1, appStartTime=1537543884622, yarnAppState=FINISHED, distributedFinalState=SUCCEEDED, appTrackingUrl=http://princeton0.field.hortonworks.com:8088/proxy/application_1536697796040_0022/, appUser=root 18/09/21 15:31:33 INFO distributedshell.Client: Application has completed successfully. Breaking monitoring loop 18/09/21 15:31:33 INFO distributedshell.Client: Application completed successfully Results: https://github.com/tspannhw/ApacheDeepLearning101/blob/master/run.log Script: https://github.com/tspannhw/ApacheDeepLearning101/blob/master/yarn.sh yarn jar /usr/hdp/current/hadoop-yarn-client/hadoop-yarn-applications-distributedshell.jar -jar /usr/hdp/current/hadoop-yarn-client/hadoop-yarn-applications-distributedshell.jar -shell_command python3.6 -shell_args "/opt/demo/ApacheDeepLearning101/analyzex.py /opt/demo/images/201813161108103.jpg" -container_resources memory-mb=512,vcores=1 For pre-HDP 3.0, see my older script using the DMLC YARN runner. We don't need that anymore. No Spark either. https://github.com/tspannhw/nifi-mxnet-yarn Python MXNet Script: https://github.com/tspannhw/ApacheDeepLearning101/blob/master/analyzehdfs.py Since we are distributed, let's write the results to HDFS. We can use and install the Python HDFS library that works on Python 2.7 and 3.x. So let's pip install it. pip install hdfs In our code: from hdfs import InsecureClient client = InsecureClient('http://princeton0.field.hortonworks.com:50070', user='root') from json import dumps client.write('/mxnetyarn/' + uniqueid + '.json', dumps(row)) We write our row as JSON to HDFS. When the job completes in YARN, we get a new JSON file written to HDFS. hdfs dfs -ls /mxnetyarn Found 2 items -rw-r--r-- 3 root hdfs 424 2018-09-21 17:50 /mxnetyarn/mxnet_uuid_img_20180921175007.json -rw-r--r-- 3 root hdfs 424 2018-09-21 17:55 /mxnetyarn/mxnet_uuid_img_20180921175552.json hdfs dfs -cat /mxnetyarn/mxnet_uuid_img_20180921175552.json {"uuid": "mxnet_uuid_img_20180921175552", "top1pct": "49.799999594688416", "top1": "n03063599 coffee mug", "top2pct": "21.50000035762787", "top2": "n07930864 cup", "top3pct": "12.399999797344208", "top3": "n07920052 espresso", "top4pct": "7.500000298023224", "top4": "n07584110 consomme", "top5pct": "5.200000107288361", "top5": "n04263257 soup bowl", "imagefilename": "/opt/demo/images/201813161108103.jpg", "runtime": "0"} HDP Assemblies https://github.com/hortonworks/hdp-assemblies/ https://github.com/hortonworks/hdp-assemblies/blob/master/tensorflow/markdown/Dockerfile.md https://github.com/hortonworks/hdp-assemblies/blob/master/tensorflow/markdown/TensorflowOnYarnTutorial.md https://github.com/hortonworks/hdp-assemblies/blob/master/tensorflow/markdown/RunTensorflowJobUsingHelperScript.md *** SUBMARINE ** Coming soon, Submarine is really cool new way. https://github.com/leftnoteasy/hadoop-1/tree/submarine/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-submarine See this awesome presentation from Strata NYC 2018 by Wangda Tan (Hortonworks): https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68289 See the quick start for setting Docker and GPU options: https://github.com/leftnoteasy/hadoop-1/blob/submarine/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-applications/hadoop-yarn-submarine/src/site/QuickStart.md Resources: https://community.hortonworks.com/articles/60480/using-images-stored-in-hdfs-for-web-pages.html ... View more

IoT Edge Processing with Apache NiFi and MiniFi and Multiple Deep Learning Libraries Series For: https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68140 See Part 1: https://community.hortonworks.com/articles/215079/iot-edge-processing-with-deep-learning-on-hdf-32-a.html See Part 2: https://community.hortonworks.com/articles/215258/iot-edge-processing-with-deep-learning-on-hdf-32-a-1.html See Part 3: https://community.hortonworks.com/articles/215271/iot-edge-processing-with-deep-learning-on-hdf-32-a-2.html You will notice a bit of a travel theme this article, it's because some of the images and work was done while on various holidays in August and September. Deep Learning We are running TensorFlow 1.10, Apache MXNet 1.3, NCSDK 2.05 and Neural Compute Application Zoo (NC App Zoo). Device Type 1: Plain Raspberry Pi (Found some old Kodak slides...) Main things to do is upgrade to Python 3.6, upgrade Raspberry PI to Stretch, upgrade libraries and a few reboots. Install OpenCV (or upgrade) and install Apache MXNet. You want to make sure you are on the latest version of Stretch and everything is cleaned up. Example: sudo apt-get upgrade sudo apt-get install build-essential tk-dev libncurses5-dev libncursesw5-dev libreadline6-dev libdb5.3-dev libgdbm-dev libsqlite3-dev libssl-dev libbz2-dev libexpat1-dev liblzma-dev zlib1g-dev sudo apt autoremove pip3.6 install --upgrade pip pip3.6 install mxnet git clone https://github.com/apache/incubator-mxnet.git --recursive Device Type 2: Raspberry Pi Enhanced with Movidius Neural Compute Stick I have updated the code to work with the new Movidius NCSDK 2.05. See: https://github.com/tspannhw/StrataNYC2018/blob/master/all2.py I also updated some variable formatting and added some additional values. Evolve that schema! So you can see some additional data: {"uuid": "mxnet_uuid_json_20180911021437.json", "label3": "n04081281 restaurant, eating house, eating place, eatery", "label1": "n03179701 desk", "roll": 4.0, "y": 0.0, "value5": "3.5%", "ipaddress": "192.168.1.156", "top5": "n03637318 lampshade, lamp shade", "label5": "n02788148 bannister, banister, balustrade, balusters, handrail", "host": "sensehatmovidius", "cputemp": 53, "top3pct": "6.5%", "diskfree": "5289.1 MB", "pressure": 1018.6, "cafferuntime": "111.685844ms", "label4": "n04009552 projector", "top4": "n03742115 medicine chest, medicine cabinet", "humidity": 42.5, "cputemp2": 52.62, "value2": "6.1%", "value3": "6.0%", "top2pct": "6.9%", "top1": "n02788148 bannister, banister, balustrade, balusters, handrail", "top4pct": "6.4%", "currenttime": "2018-09-11 02:14:44", "label2": "n03924679 photocopier", "top1pct": "7.3%", "top3": "n04286575 spotlight, spot", "starttime": "2018-09-11 02:14:33", "top5pct": "3.9%", "memory": 35.2, "value4": "5.0%", "top2": "n03250847 drumstick", "runtime": "11", "z": 1.0, "pitch": 360.0, "imagefilename": "/opt/demo/images/2018-09-10_2214.jpg", "tempf": 75.25, "temp": 35.14, "yaw": 86.0, "value1": "8.5%", "x": 0.0} Apache NiFi and MiniFi Process, Proxy, Access, Filter and Transform Data Anywhere, Anytime, Any Platform Apache NiFi and minifi Works in Moab Utah Resources: https://github.com/tspannhw/StrataNYC2018 https://www.geomesa.org/documentation/current/tutorials/geomesa-quickstart-nifi.html https://github.com/cinci/rpi-sense-hat-java https://movidius.github.io/ncsdk/install.html https://movidius.github.io/ncsdk/tf_modelzoo.html https://github.com/movidius/ncappzoo/ https://github.com/movidius/ncappzoo/blob/ncsdk2/tensorflow/facenet/README.md https://github.com/movidius/ncappzoo/blob/ncsdk2/tensorflow/inception_v4/README.md https://medium.com/tensorflow/tensorflow-1-9-officially-supports-the-raspberry-pi-b91669b0aa0 https://github.com/lhelontra/tensorflow-on-arm/releases/download/v1.10.0/tensorflow-1.10.0-cp35-none-linux_armv7l.whl https://github.com/movidius/ncappzoo/blob/ncsdk2/apps/image-classifier/README.md ... View more

IoT Edge Processing with Apache NiFi and MiniFi and Multiple Deep Learning Libraries Series For: https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68140 See Part 1: https://community.hortonworks.com/articles/215079/iot-edge-processing-with-deep-learning-on-hdf-32-a.html See Part 2: https://community.hortonworks.com/articles/215258/iot-edge-processing-with-deep-learning-on-hdf-32-a-1.html Hive - SQL - IoT Data Storage In this section, we will focus on converting JSON to AVRO to Apache ORC and storage options in Apache Hive 3. I am doing two styles of storage for one of the tables, rainbow. I am storing ORC files with an external table as well as using the Streaming API to store into an ACID table. NiFi - SQL - On Streams - Calcite SELECT * FROM FLOWFILE WHERE CAST(memory AS FLOAT) > 0 SELECT * FROM FLOWFILE WHERE CAST(tempf AS FLOAT) > 65 I check the flows as they are ingested real-time and filter based on conditions such as memory or temperature. This makes for some powerful and easy simple event processing. This is very handy when you may want to filter out standard conditions where no anomaly has occurred. IoT Data Storage Options For time series data, we are blessed with many options in HDP 3.x. The simplest choice I am doing first here. That's a simple Apache Hive 3.x table. This is where we have some tough decisions which engine to use. Hive has the best, most complete SQL and lots of interfaces. This is my default choice for where and how to store my data. If it was more than a few thousand rows a second and has a timestamp then we have to think about the architecture. Apache Druid has a lot of amazing abilities with time series data like what's coming out of these IoT devices. Since we can join Hive and Druid data and put Hive tables on top of Druid, we really should consider using Druid for our storage handler. https://cwiki.apache.org/confluence/display/Hive/Druid+Integration https://cwiki.apache.org/confluence/display/Hive/StorageHandlers https://hortonworks.com/blog/apache-hive-druid-part-1-3/ https://github.com/apache/hive/blob/master/druid-handler/src/java/org/apache/hadoop/hive/druid/DruidStorageHandlerUtils.java https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/using-druid/content/druid_anatomy_of_hive_to_druid.html We could create a Hive table backed by Druid thusly: CREATE TABLE rainbow_druid STOREDBY 'org.apache.hadoop.hive. druid.DruidStorageHandler' TBLPROPERTIES ( "druid.segment.granularity" = "MONTH", "druid.query.granularity" = "DAY") AS SELECT ts as`__time`, cast(tempf as string) s_tempf, ipaddress, cast(altitude as string) s_altitude, host, diskfree, FROM RAINBOW; For second and sub-second data, we need to consider either Druid or HBase. The nice thing is these NoSQL options also have SQL interfaces to use. It comes down to how you are going to query the data and which one you like. HBase + Phoenix is performant and been used in production forever. With HBase 2.x there are really impressive updates that make this a good option. For richer analytics and some really cool analytics with Apache Superset, it's hard not to recommend Druid. Apache Druid has really been improved recently and well integrated with Hive 3's rich querying. Example of Our Geo Data {"speed": "0.145", "diskfree": "4643.2 MB", "altitude": "6.2", "ts": "2018-08-30 17:47:03", "cputemp": 52.0, "latitude": "38.9789405", "track": "0.0", "memory": 26.5, "host": "rainbow", "uniqueid": "gps_uuid_20180830174705", "ipaddress": "172.20.10.8", "epd": "nan", "utc": "2018-08-30T17:47:05.000Z", "epx": "21.91", "epy": "31.536", "epv": "73.37", "ept": "0.005", "eps": "63.07", "longitude": "-74.824475167", "mode": "3", "time": "1535651225.0", "climb": "0.0", "epc": "nan"} Hive 3 Tables CREATE EXTERNAL TABLE IF NOT EXISTS rainbow (tempf DOUBLE, cputemp DOUBLE, pressure DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, temp DOUBLE, diskfree STRING, altitude DOUBLE, ts STRING, tempf2 DOUBLE, memory DOUBLE) STORED AS ORC LOCATION '/rainbow' create table rainbowacid(tempf DOUBLE, cputemp DOUBLE, pressure DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, temp DOUBLE, diskfree STRING, altitude DOUBLE, ts STRING, tempf2 DOUBLE, memory DOUBLE) STORED AS ORC TBLPROPERTIES ('transactional'='true') CREATE EXTERNAL TABLE IF NOT EXISTS gps (speed STRING, diskfree STRING, altitude STRING, ts STRING, cputemp DOUBLE, latitude STRING, track STRING, memory DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, epd STRING, utc STRING, epx STRING, epy STRING, epv STRING, ept STRING, eps STRING, longitude STRING, mode STRING, time STRING, climb STRING, epc STRING) STORED AS ORC LOCATION '/gps' CREATE TABLE IF NOT EXISTS gpsacid (speed STRING, diskfree STRING, altitude STRING, ts STRING, cputemp DOUBLE, latitude STRING, track STRING, memory DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, epd STRING, utc STRING, epx STRING, epy STRING, epv STRING, ept STRING, eps STRING, longitude STRING, mode STRING, `time` STRING, climb STRING, epc STRING) STORED AS ORC TBLPROPERTIES ('transactional'='true') CREATE EXTERNAL TABLE IF NOT EXISTS movidiussense (label5 STRING, runtime STRING, label1 STRING, diskfree STRING, top1 STRING, starttime STRING, label2 STRING, label3 STRING, top3pct STRING, host STRING, top5pct STRING, humidity DOUBLE, currenttime STRING, roll DOUBLE, uuid STRING, label4 STRING, tempf DOUBLE, y DOUBLE, top4pct STRING, cputemp2 DOUBLE, top5 STRING, top2pct STRING, ipaddress STRING, cputemp INT, pitch DOUBLE, x DOUBLE, z DOUBLE, yaw DOUBLE, pressure DOUBLE, top3 STRING, temp DOUBLE, memory DOUBLE, top4 STRING, imagefilename STRING, top1pct STRING, top2 STRING) STORED AS ORC LOCATION '/movidiussense' CREATE EXTERNAL TABLE IF NOT EXISTS minitensorflow2 (image STRING, ts STRING, host STRING, score STRING, human_string STRING, node_id INT) STORED AS ORC LOCATION '/minifitensorflow2' Resources: https://github.com/tspannhw/StrataNYC2018 https://www.geomesa.org/documentation/current/tutorials/geomesa-quickstart-nifi.html https://github.com/cinci/rpi-sense-hat-java ... View more

IoT Edge Processing with Deep Learning on HDF 3.2 and HDP 3.0 - Part 2 For: https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68140 See Pre-Work: https://community.hortonworks.com/articles/203638/ingesting-multiple-iot-devices-with-apache-nifi-17.html See Part 1: https://community.hortonworks.com/articles/215079/iot-edge-processing-with-deep-learning-on-hdf-32-a.html Step By Step Processing Step 1: Install Apache NiFi (One or More Nodes or clusters) Choose: https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.2.0/installing-hdf/content/install-ambari.html or docker pull hortonworks/nifi Apache NiFi Configuration for IoT https://community.hortonworks.com/articles/67756/ingesting-log-data-using-minifi-nifi.html You will need to set: nifi.remote.input.host and nifi.remote.input.socket.port in the conf/nifi.properties or Ambari settings. Step 2: Install Apache NiFi - MiniFi on Your Device(s) Download MiniFi (https://nifi.apache.org/minifi/download.html) You can choose Java or C++. For your first usage, I recommend the Java edition unless your device is too small. You can also install on a RHEL or Debian Linux machine or OSX. Download MiniFi Toolkit (https://nifi.apache.org/minifi/minifi-toolkit.html) Resources: https://cwiki.apache.org/confluence/display/MINIFI/Release+Notes#ReleaseNotes-Versioncpp-0.5.0 https://cwiki.apache.org/confluence/display/MINIFI/Release+Notes#ReleaseNotes-Version0.5.0 https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.1.2/bk_release-notes/content/ch_hdf_relnotes.html#centos7 https://community.hortonworks.com/articles/108947/minifi-for-ble-bluetooth-low-energy-beacon-data-in.html https://community.hortonworks.com/content/kbentry/107379/minifi-for-image-capture-and-ingestion-from-raspbe.html Step 3: Install Apache MXNet (On MiniFi Devices and NiFi Nodes - optional) https://mxnet.incubator.apache.org/install/index.html?platform=Devices&language=Python&processor=CPU Install build tools and build from scratch Walk through install: https://community.hortonworks.com/articles/176932/apache-deep-learning-101-using-apache-mxnet-on-the.html Resources and Source https://github.com/tspannhw/StrataNYC2018 rainbow-processing.xml rainbow-gateway-processing.xml display-images-server.xml rainbowminifi.xml https://community.hortonworks.com/articles/203638/ingesting-multiple-iot-devices-with-apache-nifi-17.html Resources and Source https://github.com/tspannhw/StrataNYC2018 rainbow-processing.xml rainbow-gateway-processing.xml display-images-server.xml rainbowminifi.xml ... View more

IoT Edge Processing with Apache NiFi and MiniFi and Multiple Deep Learning Libraries Series For: https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68140 In preparation for my talk on utilizing edge devices for deep learning, IoT sensor reading and big data processing I have updated my environment to the latest and greatest tools available. With the upgrade of HDF to 3.2, I can now use Apache NiFi 1.7 and MiniFi 0.5 for IoT data ingestion, simple event processing, conversion, data processing, data flow and storage. The architecture diagram above shows the basic flow we are utilizing. IoT Step by Step Raspberry Pi with latest patches, Python, GPS software, USB Camera, Sensor libraries, Java 8, MiniFi 0.5, TensorFlow and Apache MXNet installed. minifi flow pushes JSON and JPEGs over HTTP(s) / Site-to-Site to an Apache NiFi gateway server. Option: NiFi can push to a central NiFi cloud cluster and/or Kafka cluster both of which running on HDF 3.2 environments. Apache NiFi cluster pushes to Hive, HDFS, Dockerized API running in HDP 3.0 and Third Party APIs. NiFi and Kafka integrate with Schema Registry for our tabular data including rainbow and gps JSON data. SQL Tables in Hive I stream my data into Apache ORC files stored on HDP 3.0 HDFS directories and build external tables on them. CREATE EXTERNAL TABLE IF NOT EXISTS rainbow (tempf DOUBLE, cputemp DOUBLE, pressure DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, temp DOUBLE, diskfree STRING, altitude DOUBLE, ts STRING, tempf2 DOUBLE, memory DOUBLE) STORED AS ORC LOCATION '/rainbow'; CREATE EXTERNAL TABLE IF NOT EXISTS gps (speed STRING, diskfree STRING, altitude STRING, ts STRING, cputemp DOUBLE, latitude STRING, track STRING, memory DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, epd STRING, utc STRING, epx STRING, epy STRING, epv STRING, ept STRING, eps STRING, longitude STRING, mode STRING, time STRING, climb STRING, epc STRING) STORED AS ORC LOCATION '/gps'; For my processing needs I also have a Hive 3 ACID table for general table usage and updates. create table rainbowacid(tempf DOUBLE, cputemp DOUBLE, pressure DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, temp DOUBLE, diskfree STRING, altitude DOUBLE, ts STRING, tempf2 DOUBLE, memory DOUBLE) STORED AS ORC TBLPROPERTIES ('transactional'='true'); CREATE TABLE IF NOT EXISTS gpsacid (speed STRING, diskfree STRING, altitude STRING, ts STRING, cputemp DOUBLE, latitude STRING, track STRING, memory DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, epd STRING, utc STRING, epx STRING, epy STRING, epv STRING, ept STRING, eps STRING, longitude STRING, mode STRING, time STRING, climb STRING, epc STRING) STORED AS ORC TBLPROPERTIES ('transactional'='true'); Then I load my initial data. insert into rainbowacid select * from rainbow; insert into gpsacid select * from gps; Hive 3.x Updates %jdbc(hive) CREATE TABLE Persons_default ( ID Int NOT NULL, Name String NOT NULL, Age Int, Creator String DEFAULT CURRENT_USER(), CreateDate Date DEFAULT CURRENT_DATE() ) One of the cool new features in Hive is that you can now have defaults, as you can see which are helpful for things like standard defaults you might want like current data. This gives us even more relational style features in Hive. Another very interesting feature is materialized views which help you for having clean and fast subqueries. Here is a cool example: CREATE MATERIALIZED VIEW mv1 AS SELECT dest,origin,count(*) FROM flights_hdfs GROUP BY dest,origin References: https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/hive-overview/content/hive_whats_new_in_this_release_hive.html https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/using-hiveql/content/hive_3_internals.html https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/hive-overview/content/hive-apache-hive-3-architecturural-overview.html https://docs.hortonworks.com/HDPDocuments/HDP3/HDP-3.0.0/materialized-view/content/hive_create_materialized_view.html ... View more

Topic: IoT Edge Processing with Apache NiFi and MiniFi and Multiple Deep Learning Libraries Part 1: Multiple Devices with Data Keywords: Deep Learning On The Edge, GPS Ingestion, Sense-Hat and Rainbow Hat Sensor Ingest, WebCam Image Ingest In preparation for my talk at Strata in NYC, I am updating my IoT demos for more devices, more data types and more actions. I have three streams coming from each device including web camera images. When we are sending data from a MiniFi agent we need to define a port on an Apache NiFi server/cluster to receive it. So I design my MiniFi flow in the Apache NiFi UI (pretty soon there will be a special designer for this). You then highlight everything there and hit Create Template. You can then export it and convert to config.yml. Again, this process will be automated and connected with the NiFi Registry very shortly to make this less clicks. This is an example. When you connect to it in your flow you design in Apache NiFi UI you will connect to this port on the Remote Processor Group. If you are manually editing one (okay never do this, but sometimes I have to). You can copy that ID from this Port Details and past it in the file. Once MiniFi has it's config.yml and it's started, we will start getting messages to that Port. You can see I have two inputs, one for Movidius and one for Rainbow. I could just have one and route to what I want. It's up to you how you want to segment these flows. Welcome to Apache NiFi registry v0.2.0, this one works just as well. Very stable, but with some new magic. You can now connect to Git and Github!!!! We have structured JSON, so let's Infer a schema, clean it up and store it in the Hortonworks Schema Registry. That will make it versioned and REST enabled. I add one for the each of the two JSON file types I am sending from the rainbow device. You can see the schemas in full at the bottom of the article. The data is received from MiniFi on my local NiFi edge server for simple event processing, filtering and analysis. I route based on the two types of files, apply their schema, do a simple filter via SQL and send the converted AVRO formatted file to my cloud hosted cluster. Once I get the data I send it from my edge server to my cloud HDF 3.2 cluster. For images, I send them to my existing image storage processor group. For my other two types of files I convert them to Apache ORC and store in HDFS as Apache Hive tables. Server Dashboard Rainbow Processing Routing is Easy On High Humidity, Send a Slack Message (Query on humidity value) We can dive into any flowfile as it travels through the system and examine it's data and metadata. Now that my data is saved in HDFS with Hive tables on top I can use the latest version of Apache Zeppelin to analyze the data. I added some maps to Zeppelin via Helium, which is now available in HDP 3.0. I found a bunch of new chart types, this one could be insightful. So with the latest NiFi 1.7.1 and HDP 3.0 I can do a lot of interesting things. Next up, let's run some Dockerized TensorFlow application in my HDP 3.0 cluster. Strata Talk: https://conferences.oreilly.com/strata/strata-ny/public/schedule/detail/68140 Python Scripts https://github.com/tspannhw/StrataNYC2018/tree/master Schemas rainbow { "type": "record", "name": "rainbow", "fields": [ { "name": "tempf", "type": "double", "doc": "Type inferred from '84.15'" }, { "name": "cputemp", "type": "double", "doc": "Type inferred from '53.0'" }, { "name": "pressure", "type": "double", "doc": "Type inferred from '101028.56'" }, { "name": "host", "type": "string", "doc": "Type inferred from '\"rainbow\"'" }, { "name": "uniqueid", "type": "string", "doc": "Type inferred from '\"rainbow_uuid_20180718234222\"'" }, { "name": "ipaddress", "type": "string", "doc": "Type inferred from '\"192.168.1.165\"'" }, { "name": "temp", "type": "double", "doc": "Type inferred from '38.58'" }, { "name": "diskfree", "type": "string", "doc": "Type inferred from '\"4831.2 MB\"'" }, { "name": "altitude", "type": "double", "doc": "Type inferred from '80.65'" }, { "name": "ts", "type": "string", "doc": "Type inferred from '\"2018-07-18 23:42:22\"'" }, { "name": "tempf2", "type": "double", "doc": "Type inferred from '28.97'" }, { "name": "memory", "type": "double", "doc": "Type inferred from '32.3'" } ] } gps { "type": "record", "name": "gps", "fields": [ { "name": "speed", "type": "string", "doc": "Type inferred from '\"0.066\"'" }, { "name": "diskfree", "type": "string", "doc": "Type inferred from '\"4830.3 MB\"'" }, { "name": "altitude", "type": "string", "doc": "Type inferred from '\"43.0\"'" }, { "name": "ts", "type": "string", "doc": "Type inferred from '\"2018-07-18 23:46:39\"'" }, { "name": "cputemp", "type": "double", "doc": "Type inferred from '54.0'" }, { "name": "latitude", "type": "string", "doc": "Type inferred from '\"40.2681555\"'" }, { "name": "track", "type": "string", "doc": "Type inferred from '\"0.0\"'" }, { "name": "memory", "type": "double", "doc": "Type inferred from '32.3'" }, { "name": "host", "type": "string", "doc": "Type inferred from '\"rainbow\"'" }, { "name": "uniqueid", "type": "string", "doc": "Type inferred from '\"gps_uuid_20180718234640\"'" }, { "name": "ipaddress", "type": "string", "doc": "Type inferred from '\"192.168.1.165\"'" }, { "name": "epd", "type": "string", "doc": "Type inferred from '\"nan\"'" }, { "name": "utc", "type": "string", "doc": "Type inferred from '\"2018-07-18T23:46:40.000Z\"'" }, { "name": "epx", "type": "string", "doc": "Type inferred from '\"40.135\"'" }, { "name": "epy", "type": "string", "doc": "Type inferred from '\"42.783\"'" }, { "name": "epv", "type": "string", "doc": "Type inferred from '\"171.35\"'" }, { "name": "ept", "type": "string", "doc": "Type inferred from '\"0.005\"'" }, { "name": "eps", "type": "string", "doc": "Type inferred from '\"85.57\"'" }, { "name": "longitude", "type": "string", "doc": "Type inferred from '\"-74.529094\"'" }, { "name": "mode", "type": "string", "doc": "Type inferred from '\"3\"'" }, { "name": "time", "type": "string", "doc": "Type inferred from '\"2018-07-18T23:46:40.000Z\"'" }, { "name": "climb", "type": "string", "doc": "Type inferred from '\"0.0\"'" }, { "name": "epc", "type": "string", "doc": "Type inferred from '\"nan\"'" } ] } SQL %sql CREATE EXTERNAL TABLE IF NOT EXISTS movidiussense (label5 STRING, runtime STRING, label1 STRING, diskfree STRING, top1 STRING, starttime STRING, label2 STRING, label3 STRING, top3pct STRING, host STRING, top5pct STRING, humidity DOUBLE, currenttime STRING, roll DOUBLE, uuid STRING, label4 STRING, tempf DOUBLE, y DOUBLE, top4pct STRING, cputemp2 DOUBLE, top5 STRING, top2pct STRING, ipaddress STRING, cputemp INT, pitch DOUBLE, x DOUBLE, z DOUBLE, yaw DOUBLE, pressure DOUBLE, top3 STRING, temp DOUBLE, memory DOUBLE, top4 STRING, imagefilename STRING, top1pct STRING, top2 STRING) STORED AS ORC LOCATION '/movidiussense' %sql CREATE EXTERNAL TABLE IF NOT EXISTS minitensorflow2 (image STRING, ts STRING, host STRING, score STRING, human_string STRING, node_id INT) STORED AS ORC LOCATION '/minifitensorflow2' %sql CREATE EXTERNAL TABLE IF NOT EXISTS gps (speed STRING, diskfree STRING, altitude STRING, ts STRING, cputemp DOUBLE, latitude STRING, track STRING, memory DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, epd STRING, utc STRING, epx STRING, epy STRING, epv STRING, ept STRING, eps STRING, longitude STRING, mode STRING, time STRING, climb STRING, epc STRING) STORED AS ORC LOCATION '/gps' %sql CREATE EXTERNAL TABLE IF NOT EXISTS rainbow (tempf DOUBLE, cputemp DOUBLE, pressure DOUBLE, host STRING, uniqueid STRING, ipaddress STRING, temp DOUBLE, diskfree STRING, altitude DOUBLE, ts STRING, tempf2 DOUBLE, memory DOUBLE) STORED AS ORC LOCATION '/rainbow' References https://community.hortonworks.com/articles/176932/apache-deep-learning-101-using-apache-mxnet-on-the.html https://cwiki.apache.org/confluence/display/MINIFI/Release+Notes#ReleaseNotes-Versioncpp-0.5.0 https://cwiki.apache.org/confluence/display/MINIFI/Release+Notes#ReleaseNotes-Version0.5.0 https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.1.2/bk_release-notes/content/ch_hdf_relnotes.html#centos7 https://community.hortonworks.com/articles/108947/minifi-for-ble-bluetooth-low-energy-beacon-data-in.html https://community.hortonworks.com/content/kbentry/107379/minifi-for-image-capture-and-ingestion-from-raspbe.html NiFi Flows rainbow-server-processing.xml rainbow-minifi-ingest-in-nifi.xml ... View more

Scanning Documents into Data Lakes via Tesseract, Python, OpenCV and Apache NiFi Source: https://github.com/tspannhw/nifi-tesseract-python There are many awesome open source tools available to integrate with your Big Data Streaming flows. Take a look at these articles for installation and why the new version of Tesseract is different. I am officially recommending Python 3.6 or newer. Please don't use Python 2.7 if you don't have to. Friends don't let friends use old Python. Tesseract 4 with Deep Learning https://www.learnopencv.com/deep-learning-based-text-recognition-ocr-using-tesseract-and-opencv/ Github: https://github.com/spmallick/learnopencv/tree/master/OCR For installation on a Mac Laptop: brew install tesseract --HEAD pip3.6 install pytesseract brew install leptonica Note: if you have tesseract already, you may need to uninstall and unlink it first with brew. If you don't use brew, you can install another way. Summary Execute the run.sh (https://github.com/tspannhw/nifi-tesseract-python/blob/master/pytesstest.py) . It will send a MQTT message of the text and some other attributes in JSON format to the tesseract topic in the specified MQTT broker. Apache NiFi will read from this topic via ConsumeMQTT The flow checks to see if it's valid JSON via RouteOnContent. We run MergeRecord to convert a bunch of JSON into one big Apache Avro File Then we run ConvertAvroToORC to make a superfast Apache ORC file for storage Then we store it in HDFS via PutHDFS Running The Python Script You could have this also hooked up to a scanner or point it at a directory. You could also have it scheduled to run every 30 seconds or so. I had this hooked up to a local Apache NiFi instance to schedule runs. This can also be run by MiniFi Java Agent or MiniFi C++ agent. Or on demand if you wish. Sending MQTT Messages From Python # MQTT client = mqtt.Client() client.username_pw_set("user","pass") client.connect("server.server.com", 17769, 60) client.publish("tesseract", payload=json_string, qos=0, retain=True) You will need to run: pip3 install paho-mqtt Create the HDFS Directory hdfs dfs -mkdir -p /tesseract Create the External Hive Table (DDL Built by NiFi) CREATE EXTERNAL TABLE IF NOT EXISTS tesseract (`text` STRING, imgname STRING, host STRING, `end` STRING, te STRING, battery INT, systemtime STRING, cpu DOUBLE, diskusage STRING, memory DOUBLE, id STRING) STORED AS ORC LOCATION '/tesseract'; This DDL is a side effect, it's built by our ORC conversion and HDFS storage commands. You could run that create script in Hive View 2, Beeline or another Apache Hive JDBC/ODBC tool. I used Apache Zeppelin since I am going to be doing queries there anyway. Let's Ingest Our Captured Images and Process Them with Apache Tika, TensorFlow and grab the metadata Consume MQTT Records and Store in Apache Hive Let's look at other fields in Zeppelin Let's Look at Our Records in Apache Zeppelin via a SQL Query (SELECT *FROM TESSERACT) ConsumeMQTT: Give me all the record from the tesseract topic from our MQTT Broker. Isolation from our ingest clients which could be 100,000 devices. MergeRecord: Merge all the JSON files sent via MQTT into one big AVRO File ConvertAVROToORC: converts are merged AVRO file PutHDFS Tesseract Example Schema in Hortonworks Schema Registry TIP: You can generate your schema with InferAvroSchema. Do that once, copy it and paste into Schema Registry. Then you can remove that step from your flow. The Schema Text { "type": "record", "name": "tesseract", "fields": [ { "name": "text", "type": "string", "doc": "Type inferred from '\"cgi cctong aiternacrety, pou can acces the complete Pro\nLance repesiiry from eh Provenance mens: The Provenance\n‘emu inchades the Date/Time, Actontype, the Unsque Fowie\nTD and other sata. Om the ar it is smal exci i oe:\n‘ick chs icon, and you get the flowin On the right, war\n‘cots like three inthe cic soemecaed gether Liege:\n\nLineage ts visualined as « lange direcnad sqycie graph (DAG) char\nSrones the seeps 1m she Gow where modifications oF routing ‘oot\nplace on the Aewiike. Righe-iieit « step lp the Lineage s view\nSetusls aboot the fowtle at that step ar expand the ow to ander:\nScand where & was potentially domed frum. Af the very bottom\nleft of the Lineage Oi a slider wath a play button to play the pro\n“sing flow (with scaled ame} and understand where tbe owtise\nSpent the meat Game of at whch PORN get muted\n\naide the Bowtie dealin, you cam: finn deed analy of box\n\ntern\n=\"'" }, { "name": "imgname", "type": "string", "doc": "Type inferred from '\"images/tesseract_image_20180613205132_c14779b8-1546-433e-8976-ddb5bfc5f978.jpg\"'" }, { "name": "host", "type": "string", "doc": "Type inferred from '\"HW13125.local\"'" }, { "name": "end", "type": "string", "doc": "Type inferred from '\"1528923095.3205361\"'" }, { "name": "te", "type": "string", "doc": "Type inferred from '\"3.7366552352905273\"'" }, { "name": "battery", "type": "int", "doc": "Type inferred from '100'" }, { "name": "systemtime", "type": "string", "doc": "Type inferred from '\"06/13/2018 16:51:35\"'" }, { "name": "cpu", "type": "double", "doc": "Type inferred from '22.8'" }, { "name": "diskusage", "type": "string", "doc": "Type inferred from '\"113759.7 MB\"'" }, { "name": "memory", "type": "double", "doc": "Type inferred from '69.4'" }, { "name": "id", "type": "string", "doc": "Type inferred from '\"20180613205132_c14779b8-1546-433e-8976-ddb5bfc5f978\"'" } ] } The above schema was generated by Infer Avro Schema in Apache NiFi. Image Analytics Results { "tiffImageWidth" : "1280", "ContentType" : "image/jpeg", "JPEGImageWidth" : "1280 pixels", "FileTypeDetectedFileTypeName" : "JPEG", "tiffBitsPerSample" : "8", "ThumbnailHeightPixels" : "0", "label4" : "book jacket", "YResolution" : "1 dot", "label5" : "pill bottle", "ImageWidth" : "1280 pixels", "JFIFYResolution" : "1 dot", "JPEGImageHeight" : "720 pixels", "filecreationTime" : "2018-06-13T17:24:07-0400", "JFIFThumbnailHeightPixels" : "0", "DataPrecision" : "8 bits", "XResolution" : "1 dot", "ImageHeight" : "720 pixels", "JPEGNumberofComponents" : "3", "JFIFXResolution" : "1 dot", "FileTypeExpectedFileNameExtension" : "jpg", "JPEGDataPrecision" : "8 bits", "FileSize" : "223716 bytes", "probability4" : "1.74%", "tiffImageLength" : "720", "probability3" : "3.29%", "probability2" : "6.13%", "probability1" : "81.23%", "FileName" : "apache-tika-2858986094088526803.tmp", "filelastAccessTime" : "2018-06-13T17:24:07-0400", "JFIFThumbnailWidthPixels" : "0", "JPEGCompressionType" : "Baseline", "JFIFVersion" : "1.1", "filesize" : "223716", "FileModifiedDate" : "Wed Jun 13 17:24:27 -04:00 2018", "Component3" : "Cr component: Quantization table 1, Sampling factors 1 horiz/1 vert", "Component1" : "Y component: Quantization table 0, Sampling factors 2 horiz/2 vert", "Component2" : "Cb component: Quantization table 1, Sampling factors 1 horiz/1 vert", "NumberofTables" : "4 Huffman tables", "FileTypeDetectedFileTypeLongName" : "Joint Photographic Experts Group", "fileowner" : "tspann", "filepermissions" : "rw-r--r--", "JPEGComponent3" : "Cr component: Quantization table 1, Sampling factors 1 horiz/1 vert", "JPEGComponent2" : "Cb component: Quantization table 1, Sampling factors 1 horiz/1 vert", "JPEGComponent1" : "Y component: Quantization table 0, Sampling factors 2 horiz/2 vert", "FileTypeDetectedMIMEType" : "image/jpeg", "NumberofComponents" : "3", "HuffmanNumberofTables" : "4 Huffman tables", "label1" : "menu", "XParsedBy" : "org.apache.tika.parser.DefaultParser, org.apache.tika.parser.ocr.TesseractOCRParser, org.apache.tika.parser.jpeg.JpegParser", "label2" : "web site", "label3" : "crossword puzzle", "absolutepath" : "/Volumes/seagate/opensourcecomputervision/images/", "filelastModifiedTime" : "2018-06-13T17:24:07-0400", "ThumbnailWidthPixels" : "0", "filegroup" : "staff", "ResolutionUnits" : "none", "JFIFResolutionUnits" : "none", "CompressionType" : "Baseline", "probability5" : "1.12%" } This is built using a combination of Apache Tika, TensorFlow and other metadata analysis processors. ... View more