Deep Speech with Apache NiFi 1.8 Tools: Python 3.6, PyAudio, TensorFlow, Deep Speech, Shell, Apache NiFi Why: Speech-to-Text Use Case: Voice control and recognition. Series: Holiday Use Case: Turn on Holiday Lights and Music on command. Cool Factor: Ever want to run a query on Live Ingested Voice Commands? Other Options: https://community.hortonworks.com/articles/155519/voice-controlled-data-flows-with-google-aiy-voice.html We are using Python 3.6 to write some code around pyaudio, tensorflow and Deep Speech to capture audio, store it in a wave file and then process it with Deep Speech to extract some text. This example is running in OSX without a GPU on Tensorflow v1.11. The Mozilla Github repo for their Deep Speech implementation has nice getting started information that I used to integrate our flow with Apache NiFi. Installation as per https://github.com/mozilla/DeepSpeech pip3 install deepspeech wget -O - https://github.com/mozilla/DeepSpeech/releases/download/v0.3.0/deepspeech-0.3.0-models.tar.gz | tar xvfz - This pre-trained model is available for English. For other languages, you will need to build your own. You can use a beef HDP 3.1 cluster to train this. Note: THIS IS A 1.8 GIG DOWNLOAD. That may be an issue for laptops, devices or small data people. Apache NiFi Flow The flow is simple, we call our shell script that runs Python that records audio and sends it to Deep Speech for processing. We get back a voice_string in JSON that we turn into a record for querying and filtering in Apache NiFi. I am handling a few voice commands for "Save", "Load" and "Move". As you can imagine you can handle pretty much anything you want. It's a simple way to use voice to control streaming data flows or just to ingest large streams of text. Even using advanced Deep Learning, text recognition is still not the strongest. If you are going to load balance connections between nodes, you have options on compression and load balancing strategies. This can come in handy if you have a lot of servers. Shell Script python3.6 /Volumes/TSPANN/projects/DeepSpeech/processnifi.py /Volumes/TSPANN/projects/DeepSpeech/models/output_graph.pbmm /Volumes/TSPANN/projects/DeepSpeech/models/alphabet.txt Schema { "type" : "record", "name" : "voice", "fields" : [ { "name" : "systemtime", "type" : "string", "doc" : "Type inferred from '\"12/10/2018 14:53:47\"'" }, { "name" : "voice_string", "type" : "string", "doc" : "Type inferred from '\"\"'" } ] } We can add more fields as needed. Example Run HW13125:DeepSpeech tspann$ ./runnifi.sh TensorFlow: v1.11.0-9-g97d851f04e DeepSpeech: unknown 2018-12-10 14:36:43.714433: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA {"systemtime": "12/10/2018 14:36:43", "voice_string": "one two three or five six seven eight nine"} We can run this on top of YARN 3.1 as dockerized or non-dockerized workloads. Setting up nodes to run HDF 3.3 - Apache NiFi and friends is easy in the cloud or on-premise in OpenStack with super devops tools. When running Apache NiFi it is easy to monitor in Ambari: References: https://github.com/mozilla/DeepSpeech https://community.hortonworks.com/articles/224268/running-tensorflow-on-yarn-31-with-or-without-gpu.html https://arxiv.org/abs/1412.5567 https://github.com/tspannhw/nifi-deepspeech ... View more

Apache NiFi Processor for Apache MXNet SSD: Single Shot MultiBox Object Detector (Deep Learning) The news is out, Apache MXNet has added a Java API. So as soon as I could I got my hands on the maven repo and an example program and got to work writing a new Apache NiFi processor for it. I have run this on standalone Apache NiFi 1.8.0 and on HDF 3.3 - Apache NiFi 1.8.0 and both work. So anyone who wants to be an alpha tester, please download it and give it a try. Apache MXNet SSD is a good example of a pretrained deep learning model that works pretty well for general images in a use cases especially around people and cars. You can fine-tune this with some more images and runs: https://mxnet.incubator.apache.org/faq/finetune.html The nice thing is now we can start including Apache MXNet as part of Java applications such as Kafka Streams, Apache Storm, Apache Spark, Spring Boot and other use cases using Java. I could potentially inject this into a Hive UDF (https://community.hortonworks.com/articles/39980/creating-a-hive-udf-in-java.html#comment-40026) or Pig UDF. The performance may be fast enough. We now have four Java options for Deep Learning: DL4J, H2O, Tensorflow and Apache MXNet. Unfortunately, both TensorFlow and MXNet Java APIs are not quite production ready. I may do some further research on running MXNet as a Hive UDF, it would be cool to have in a query. For those who don't want to setup a development environment with JDK 8+, Maven 3.3+ and git, you can download a pre-built nar file here: https://github.com/tspannhw/nifi-mxnetinference-processor/releases/tag/v1.0. As part of the recent release of HDF 3.3, I have upgraded my OpenStack Centos 7 cluster. Important Caveats Notice, the Java API is in preview and so is this processor. Do not use this in production! This is in development and I am the only one working on it. The Java API from Apache MXNet is in flux and will be changing. See the POM as it is tied to the OSX/Mac version of the library. You will need to change that. You will need to download the pre-built MXNet model and place it in a directory accessible to Apache NiFi server/cluster. I am still cleaning up the rectangle code for identifying objects in the pictures. As you will notice, my rectangle drawing is a bit off. I need to work on that. Once you drop your built nar file and models in the nifi/lib directory and restart Apache NiFi, you can add it to your canvas. We need to feed it some images. You can use my web cam processor, an image URL feed or local files. To grab images from an HTTPS site, you need an SSL Context Service like this StandardSSLContextService below. You will need to point to the cacerts used by the JRE/JDK running your Apache NiFi node. The default password in Java is changeme. Hopefully you have changed it. To configure my new processor, just put in the full path to the model directory and then "/resnet50_ssd_model" as that is the prefix for the model. Our example flow with new processor being fed by traffic cameras, webcams, local files and local webcam. Some output of our flow: Our top 5 probabilities and labels Example Data: { "ymin_1" : "456.01", "ymin_5" : "159.29", "ymin_4" : "235.83", "ymin_3" : "206.64", "ymin_2" : "383.84", "label_5" : "person", "xmax_5" : "121.14", "label_4" : "bicycle", "xmax_4" : "137.89", "label_3" : "dog", "xmax_3" : "179.14", "ymax_1" : "150.66", "ymax_2" : "418.95", "ymax_3" : "476.79", "label_2" : "bicycle", "label_1" : "car", "probability_4" : "0.22", "probability_5" : "0.13", "probability_2" : "0.90", "xmin_5" : "88.93", "probability_3" : "0.82", "ymax_4" : "413.43", "probability_1" : "1.00", "ymax_5" : "190.04", "xmax_2" : "149.96", "xmax_1" : "72.03", "xmin_3" : "83.82", "xmin_4" : "93.05", "xmin_1" : "312.21", "xmin_2" : "155.96" } Resources: https://medium.com/apache-mxnet/introducing-java-apis-for-deep-learning-inference-with-apache-mxnet-8406a698fa5a https://github.com/apache/incubator-mxnet/tree/java-api/scala-package/examples/src/main/java/org/apache/mxnetexamples/javaapi https://mxnet.incubator.apache.org/install/java_setup.html Source: https://github.com/tspannhw/nifi-mxnetinference-processor Video walk-through: https://www.youtube.com/watch?v=Q4dSGPvqXSA&t=196s&list=PL-7XqvSmQqfTSihuoIP_ZAnN7mFIHkZ_e&index=17 mxnet-processor.xml Download the artifacts listed: https://github.com/apache/incubator-mxnet/tree/java-api/scala-package/examples/src/main/java/org/apache/mxnetexamples/javaapi/infer/objectdetector#step-1 Maven POM (I used Java 8 and Maven 3.3.9) <?xml version="1.0" encoding="UTF-8"?> <!-- Licensed to the Apache Software Foundation (ASF) under one or more contributor license agreements. See the NOTICE file distributed with this work for additional information regarding copyright ownership. The ASF licenses this file to You under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. --> <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"> <modelVersion>4.0.0</modelVersion> <parent> <groupId>com.dataflowdeveloper.mxnet</groupId> <artifactId>inference</artifactId> <version>1.0</version> </parent> <artifactId>nifi-mxnetinference-processors</artifactId> <packaging>jar</packaging> <dependencies> <dependency> <groupId>org.apache.nifi</groupId> <artifactId>nifi-api</artifactId> </dependency> <dependency> <groupId>org.apache.nifi</groupId> <artifactId>nifi-utils</artifactId> <version>1.8.0</version> </dependency> <dependency> <groupId>org.apache.nifi</groupId> <artifactId>nifi-mock</artifactId> <version>1.8.0</version> <scope>test</scope> </dependency> <dependency> <groupId>org.slf4j</groupId> <artifactId>slf4j-simple</artifactId> <scope>test</scope> </dependency> <dependency> <groupId>junit</groupId> <artifactId>junit</artifactId> <scope>test</scope> </dependency> <dependency> <groupId>org.apache.mxnet</groupId> <artifactId>mxnet-full_2.11-osx-x86_64-cpu</artifactId> <version>1.3.1-SNAPSHOT</version> </dependency> </dependencies> </project> I have moved from Eclipse to IntelliJ from my builds. I am looking at Apache Netbeans as well. ... View more

MiniFi Java Agent 0.5 Copy over necessary NARs from Apache NiFi 1.7 lib: nifi-ssl-context-service-nar-1.7.0.nar nifi-standard-services-api-nar-1.7.0.nar nifi-kafka-1-0-nar-1.7.0.nar This will support PublishKafka_1_0 and ConsumeKafka_1_0. Then create a consume and/or publish flow. You can combine the two based on your needs. In my simple example I consume the Kafka messages in MiniFi and write to a file. I also write the metadata to a JSON file. Consume Kafka Publish Electric Monitoring Data To Kafka Let's monitor the messages going through our topic, smartPlug. Publish Messages to Kafka Consume Any Messages From the smartPlug topic Logs Provenance Event file containing 377 records. In the past 5 minutes, 1512 events have been written to the Provenance Repository, totaling 839.32 KB 2018-11-26 19:42:32,473 INFO [main] o.a.n.c.s.StandardProcessScheduler Starting PutFile[id=25a86505-031a-37d9-0000-000000000000]2018-11-26 19:42:32,474 INFO [main] o.a.n.c.s.StandardProcessScheduler Starting UpdateAttribute[id=9220d40d-ee1d-3f61-0000-000000000000]2018-11-26 19:42:32,474 INFO [main] o.apache.nifi.controller.FlowController Started 0 Remote Group Ports transmitting2018-11-26 19:42:32,478 INFO [main] org.apache.nifi.minifi.MiNiFiServer Flow loaded successfully.2018-11-26 19:42:32,479 INFO [Monitor Processor Lifecycle Thread-2] o.a.n.c.s.TimerDrivenSchedulingAgent Scheduled ConsumeKafka_1_0[id=8556f1ce-a915-3fda-0000-000000000000] to run with 1 threads2018-11-26 19:42:32,479 INFO [main] org.apache.nifi.BootstrapListener Successfully initiated communication with Bootstrap2018-11-26 19:42:32,479 INFO [Monitor Processor Lifecycle Thread-1] o.a.n.c.s.TimerDrivenSchedulingAgent Scheduled AttributesToJSON[id=0628b4e5-10d0-3b09-0000-000000000000] to run with 1 threads2018-11-26 19:42:32,479 INFO [main] org.apache.nifi.minifi.MiNiFi Controller initialization took 2787584123 nanoseconds.2018-11-26 19:42:32,480 INFO [Monitor Processor Lifecycle Thread-1] o.a.n.c.s.TimerDrivenSchedulingAgent Scheduled PutFile[id=25a86505-031a-37d9-0000-000000000000] to run with 1 threads2018-11-26 19:42:32,481 INFO [Monitor Processor Lifecycle Thread-2] o.a.n.c.s.TimerDrivenSchedulingAgent Scheduled UpdateAttribute[id=9220d40d-ee1d-3f61-0000-000000000000] to run with 1 threads2018-11-26 19:42:32,585 INFO [Timer-Driven Process Thread-2] o.a.k.clients.consumer.ConsumerConfig ConsumerConfig values:auto.commit.interval.ms = 5000auto.offset.reset = latestbootstrap.servers = [princeton1.field.hortonworks.com:6667]check.crcs = trueclient.id =connections.max.idle.ms = 540000enable.auto.commit = falseexclude.internal.topics = truefetch.max.bytes = 52428800fetch.max.wait.ms = 500fetch.min.bytes = 1group.id = minificonsumer1heartbeat.interval.ms = 3000interceptor.classes = nullinternal.leave.group.on.close = trueisolation.level = read_uncommittedkey.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializermax.partition.fetch.bytes = 1048576max.poll.interval.ms = 300000max.poll.records = 10000metadata.max.age.ms = 300000metric.reporters = []metrics.num.samples = 2metrics.recording.level = INFOmetrics.sample.window.ms = 30000partition.assignment.strategy = [class org.apache.kafka.clients.consumer.RangeAssignor]receive.buffer.bytes = 65536reconnect.backoff.max.ms = 1000reconnect.backoff.ms = 50request.timeout.ms = 305000retry.backoff.ms = 100sasl.jaas.config = nullsasl.kerberos.kinit.cmd = /usr/bin/kinitsasl.kerberos.min.time.before.relogin = 60000sasl.kerberos.service.name = nullsasl.kerberos.ticket.renew.jitter = 0.05sasl.kerberos.ticket.renew.window.factor = 0.8sasl.mechanism = GSSAPIsecurity.protocol = PLAINTEXTsend.buffer.bytes = 131072session.timeout.ms = 10000ssl.cipher.suites = nullssl.enabled.protocols = [TLSv1.2, TLSv1.1, TLSv1]ssl.endpoint.identification.algorithm = nullssl.key.password = nullssl.keymanager.algorithm = SunX509ssl.keystore.location = nullssl.keystore.password = nullssl.keystore.type = JKSssl.protocol = TLSssl.provider = nullssl.secure.random.implementation = nullssl.trustmanager.algorithm = PKIXssl.truststore.location = nullssl.truststore.password = nullssl.truststore.type = JKSvalue.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializer2018-11-26 19:42:32,727 INFO [Timer-Driven Process Thread-2] o.a.kafka.common.utils.AppInfoParser Kafka version : 1.0.02018-11-26 19:42:32,727 INFO [Timer-Driven Process Thread-2] o.a.kafka.common.utils.AppInfoParser Kafka commitId : aaa7af6d4a11b29d2018-11-26 19:42:33,088 INFO [Timer-Driven Process Thread-2] o.a.k.c.c.internals.AbstractCoordinator [Consumer clientId=consumer-1, groupId=minificonsumer1] Discovered coordinator princeton1.field.hortonworks.com:6667 (id: 2147482646 rack: null)2018-11-26 19:42:33,090 INFO [Timer-Driven Process Thread-2] o.a.k.c.c.internals.ConsumerCoordinator [Consumer clientId=consumer-1, groupId=minificonsumer1] Revoking previously assigned partitions []2018-11-26 19:42:33,091 INFO [Timer-Driven Process Thread-2] o.a.k.c.c.internals.AbstractCoordinator [Consumer clientId=consumer-1, groupId=minificonsumer1] (Re-)joining group2018-11-26 19:42:36,391 INFO [Timer-Driven Process Thread-2] o.a.k.c.c.internals.AbstractCoordinator [Consumer clientId=consumer-1, groupId=minificonsumer1] Successfully joined group with generation 32018-11-26 19:42:36,394 INFO [Timer-Driven Process Thread-2] o.a.k.c.c.internals.ConsumerCoordinator [Consumer clientId=consumer-1, groupId=minificonsumer1] Setting newly assigned partitions [smartPlug-0]2018-11-26 19:44:32,325 INFO [pool-34-thread-1] o.a.n.c.r.WriteAheadFlowFileRepository Successfully checkpointed FlowFile Repository with 0 records in 0 milliseconds2018-11-26 19:44:40,700 INFO [Provenance Maintenance Thread-1] o.a.n.p.PersistentProvenanceRepository Created new Provenance Event Writers for events starting with ID 14372018-11-26 19:44:40,765 INFO [Provenance Repository Rollover Thread-1] o.a.n.p.lucene.SimpleIndexManager Index Writer for provenance_repository/index-1543271506000 has been returned to Index Manager and is no longer in use. Closing Index Writer2018-11-26 19:44:40,767 INFO [Provenance Repository Rollover Thread-1] o.a.n.p.PersistentProvenanceRepository Successfully merged 16 journal files (28 records) into single Provenance Log File provenance_repository/1409.prov in 62 milliseconds2018-11-26 19:44:40,768 INFO [Provenance Repository Rollover Thread-1] o.a.n.p.PersistentProvenanceRepository Successfully Rolled over Provenance Event file containing 151 records. In the past 5 minutes, 28 events have been written to the Provenance Repository, totaling 15.43 KB JSON Kafka Message and JSON Kafka Metadata Stored As Files monitor/1448678223641638.attr.json {"path":"./","filename":"1448678223641638","kafka.partition":"0","kafka.offset":"5543","kafka.topic":"smartPlug","kafka.key":"cb90ad21-b311-494c-96cc-06dd2e8747df","uuid":"041459fc-c63e-4056-ab50-1c375cd7d49f"} monitor/1448678223641638 {"day30": 0.431, "day31": 1.15, "sw_ver": "1.2.5 Build 171206 Rel.085954", "hw_ver": "1.0", "mac": "50:C7:BF:B1:95:D5", "type": "IOT.SMARTPLUGSWITCH", "hwId": "60FF6B258734EA6880E186F8C96DDC61", "fwId": "00000000000000000000000000000000", "oemId": "FFF22CFF774A0B89F7624BFC6F50D5DE", "dev_name": "Wi-Fi Smart Plug With Energy Monitoring", "model": "HS110(US)", "deviceId": "8006ECB1D454C4428953CB2B34D9292D18A6DB0E", "alias": "Tim", "icon_hash": "", "relay_state": 1, "on_time": 886569, "active_mode": "schedule", "feature": "TIM:ENE", "updating": 0, "rssi": -75, "led_off": 0, "latitude": 40.268216, "longitude": -74.529088, "index": 18, "zone_str": "(UTC-05:00) Eastern Daylight Time (US & Canada)", "tz_str": "EST5EDT,M3.2.0,M11.1.0", "dst_offset": 60, "month10": 1.581, "month11": 30.888, "current": 0.067041, "voltage": 122.151701, "power": 1.277361, "total": 24.289, "time": "11/26/2018 21:54:22", "ledon": true, "systemtime": "11/26/2018 21:54:22"} Resources: https://blog.ona.io/general/2017/08/30/streaming-ona-data-with-nifi-kafka-druid-and-superset.html https://community.hortonworks.com/articles/193945/social-media-monitoring-with-nifi-hivedruid-integr.html https://community.hortonworks.com/articles/177561/streaming-tweets-with-nifi-kafka-tranquility-druid.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/kafka-using-kafka-streams/content/kafka-using-kafka-streams.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/minifi-quick-start/content/overview.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/minifi-quick-start/content/before_you_begin.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/minifi-quick-start/content/installing_minifi_on_linux.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/minifi-quick-start/content/using_processors_not_packaged_with_minifi.html?es_p=8055369 https://community.hortonworks.com/articles/227560/real-time-stock-processing-with-apache-nifi-and-ap.html https://docs.hortonworks.com/HDPDocuments/HDF3/HDF-3.3.0/minifi-quick-start/content/using_processors_not_packaged_with_minifi.html?es_p=8055369 Files: consumekafka2.xml pushkafka1.xml configyml consume.txt configymlsend.txt ... View more

Implementing Streaming Use Case From REST to Hive with Apache NiFi and Apache Kafka Part 1 With Apache Kafka 2.0, Apache NiFi 1.8 and many new features and abilities coming out. It's time to put them to the test. So to plan out what we are going to do, I have a high level architecture diagram. We are going to ingest a number of sources including REST feeds, Social Feeds, Messages, Images, Documents and Relational Data. We will ingest with NiFi, filter and process and segment it into Kafka topics. Kafka data will be in Apache Avro format with schemas specified in Hortonworks Schema Registry. Kafka Streams, Spark and NiFi will do additional event processing along with machine learning and deep learning. it will be stored in Druid for real-time analytics and summaries. Hive, HDFS and S3 will store for permanent storage. We will do dashboards with Superset and Spark SQL + Zeppelin. We will integrate machine learning with Spark ML, TensorFlow and Apache MXNet. We will also push back cleaned and aggregated data to subscribers via Kafka and NiFi. We will push to Dockerized applications, message listeners, web clients, Slack channels and to email mailing lists. To be useful in our enterprise, we will have full authorization, authentication, auditing, data encryption and data lineage via Apache Ranger, Apache Atlas and Apache NiFi. NiFi Registry and github will be used for source code control. We will have administration capabilities via Apache Ambari. An example server layout: NiFi Flows Real-time free stock data is available from IEX with no license key. The data streams in very fast, thankfully that's no issue for Apache NiFi and Kafka. Consume the Different Records from topics and store to HDFS in separate directories and tables. Let's split up one big REST file into individual records of interest. Our REST feed has quote, chart and news arrays. Let's Push Some Messages to Slack We can easily consume from multiple topics in Apache NiFi. Querying data is easy as it's in motion, since we have schemas We create schemas for each of our Kafka Topics We can monitor all these messages going through Kafka in Ambari (and also in much better detail in Hortonworks SMM). I read in data and then can push it to Kafka 1.0 and 2.0 brokers. Once data is sent, NiFi let's us know. Projects Used Apache Kafka Apache Kafka Streams Apache MXNet NLTK Stanford CoreNLP Apache OpenNLP TextBlob SpaCy Apache NiFi Apache Druid Apache Hive on Kafka Apache Hive on Druid Apache Hive on JDBC Apache Zeppelin NLP - Apache OpenNLP and Stanford CoreNLP Hortonworks Schema Registry NiFi Registry Apache Ambari Log Search Hortonworks SMM Hortonworks Data Plane Services (DPS) Sources REST Twitter JDBC Sensors MQTT Documents Sinks Apache Hadoop HDFS Apache Kafka Apache Hive Slack S3 Apache Druid Apache HBase Topics iextradingnews iextradingquote iextradingchart stocks cyber HDFS Directories hdfs dfs -mkdir -p /iextradingnews hdfs dfs -mkdir -p /iextradingquote hdfs dfs -mkdir -p /iextradingchart hdfs dfs -mkdir -p /stocks hdfs dfs -mkdir -p /cyber hdfs dfs -chmod -R 777 / PutHDFS /${kafka.topic} /iextradingchart/859496561256574.orc /iextradingnews/855935960267509.orc /iextradingquote/859143934804532.orc Hive Tables CREATE EXTERNAL TABLE IF NOT EXISTS iextradingchart (`date` STRING, open DOUBLE, high DOUBLE, low DOUBLE, close DOUBLE, volume INT, unadjustedVolume INT, change DOUBLE, changePercent DOUBLE, vwap DOUBLE, label STRING, changeOverTime INT) STORED AS ORC LOCATION '/iextradingchart'; CREATE EXTERNAL TABLE IF NOT EXISTS iextradingquote (symbol STRING, companyName STRING, primaryExchange STRING, sector STRING, calculationPrice STRING, open DOUBLE, openTime BIGINT, close DOUBLE, closeTime BIGINT, high DOUBLE, low DOUBLE, latestPrice DOUBLE, latestSource STRING, latestTime STRING, latestUpdate BIGINT, latestVolume INT, iexRealtimePrice DOUBLE, iexRealtimeSize INT, iexLastUpdated BIGINT, delayedPrice DOUBLE, delayedPriceTime BIGINT, extendedPrice DOUBLE, extendedChange DOUBLE, extendedChangePercent DOUBLE, extendedPriceTime BIGINT, previousClose DOUBLE, change DOUBLE, changePercent DOUBLE, iexMarketPercent DOUBLE, iexVolume INT, avgTotalVolume INT, iexBidPrice INT, iexBidSize INT, iexAskPrice INT, iexAskSize INT, marketCap INT, peRatio DOUBLE, week52High DOUBLE, week52Low DOUBLE, ytdChange DOUBLE) STORED AS ORC LOCATION '/iextradingquote'; CREATE EXTERNAL TABLE IF NOT EXISTS iextradingnews (`datetime` STRING, headline STRING, source STRING, url STRING, summary STRING, related STRING, image STRING) STORED AS ORC LOCATION '/iextradingnews'; Schemas { "type": "record", "name": "iextradingchart", "fields": [ { "name": "date", "type": [ "string", "null" ] }, { "name": "open", "type": [ "double", "null" ] }, { "name": "high", "type": [ "double", "null" ] }, { "name": "low", "type": [ "double", "null" ] }, { "name": "close", "type": [ "double", "null" ] }, { "name": "volume", "type": [ "int", "null" ] }, { "name": "unadjustedVolume", "type": [ "int", "null" ] }, { "name": "change", "type": [ "double", "null" ] }, { "name": "changePercent", "type": [ "double", "null" ] }, { "name": "vwap", "type": [ "double", "null" ] }, { "name": "label", "type": [ "string", "null" ] }, { "name": "changeOverTime", "type": [ "int", "null" ] } ]}{ "type": "record", "name": "iextradingquote", "fields": [ { "name": "symbol", "type": [ "string", "null" ], "doc": "Type inferred from '\"HDP\"'" }, { "name": "companyName", "type": [ "string", "null" ], "doc": "Type inferred from '\"Hortonworks Inc.\"'" }, { "name": "primaryExchange", "type": [ "string", "null" ], "doc": "Type inferred from '\"Nasdaq Global Select\"'" }, { "name": "sector", "type": [ "string", "null" ], "doc": "Type inferred from '\"Technology\"'" }, { "name": "calculationPrice", "type": [ "string", "null" ], "doc": "Type inferred from '\"close\"'" }, { "name": "open", "type": [ "double", "null" ], "doc": "Type inferred from '16.3'" }, { "name": "openTime", "type": [ "long", "null" ], "doc": "Type inferred from '1542033000568'" }, { "name": "close", "type": [ "double", "null" ], "doc": "Type inferred from '15.76'" }, { "name": "closeTime", "type": [ "long", "null" ], "doc": "Type inferred from '1542056400520'" }, { "name": "high", "type": [ "double", "null" ], "doc": "Type inferred from '16.37'" }, { "name": "low", "type": [ "double", "null" ], "doc": "Type inferred from '15.2'" }, { "name": "latestPrice", "type": [ "double", "null" ], "doc": "Type inferred from '15.76'" }, { "name": "latestSource", "type": [ "string", "null" ], "doc": "Type inferred from '\"Close\"'" }, { "name": "latestTime", "type": [ "string", "null" ], "doc": "Type inferred from '\"November 12, 2018\"'" }, { "name": "latestUpdate", "type": [ "long", "null" ], "doc": "Type inferred from '1542056400520'" }, { "name": "latestVolume", "type": [ "int", "null" ], "doc": "Type inferred from '4012339'" }, { "name": "iexRealtimePrice", "type": [ "double", "null" ], "doc": "Type inferred from '15.74'" }, { "name": "iexRealtimeSize", "type": [ "int", "null" ], "doc": "Type inferred from '43'" }, { "name": "iexLastUpdated", "type": [ "long", "null" ], "doc": "Type inferred from '1542056397411'" }, { "name": "delayedPrice", "type": [ "double", "null" ], "doc": "Type inferred from '15.76'" }, { "name": "delayedPriceTime", "type": [ "long", "null" ], "doc": "Type inferred from '1542056400520'" }, { "name": "extendedPrice", "type": [ "double", "null" ], "doc": "Type inferred from '15.85'" }, { "name": "extendedChange", "type": [ "double", "null" ], "doc": "Type inferred from '0.09'" }, { "name": "extendedChangePercent", "type": [ "double", "null" ], "doc": "Type inferred from '0.00571'" }, { "name": "extendedPriceTime", "type": [ "long", "null" ], "doc": "Type inferred from '1542059622726'" }, { "name": "previousClose", "type": [ "double", "null" ], "doc": "Type inferred from '16.24'" }, { "name": "change", "type": [ "double", "null" ], "doc": "Type inferred from '-0.48'" }, { "name": "changePercent", "type": [ "double", "null" ], "doc": "Type inferred from '-0.02956'" }, { "name": "iexMarketPercent", "type": [ "double", "null" ], "doc": "Type inferred from '0.03258'" }, { "name": "iexVolume", "type": [ "int", "null" ], "doc": "Type inferred from '130722'" }, { "name": "avgTotalVolume", "type": [ "int", "null" ], "doc": "Type inferred from '2042809'" }, { "name": "iexBidPrice", "type": [ "int", "null" ], "doc": "Type inferred from '0'" }, { "name": "iexBidSize", "type": [ "int", "null" ], "doc": "Type inferred from '0'" }, { "name": "iexAskPrice", "type": [ "int", "null" ], "doc": "Type inferred from '0'" }, { "name": "iexAskSize", "type": [ "int", "null" ], "doc": "Type inferred from '0'" }, { "name": "marketCap", "type": [ "int", "null" ], "doc": "Type inferred from '1317308142'" }, { "name": "peRatio", "type": [ "double", "null" ], "doc": "Type inferred from '-7.43'" }, { "name": "week52High", "type": [ "double", "null" ], "doc": "Type inferred from '26.22'" }, { "name": "week52Low", "type": [ "double", "null" ], "doc": "Type inferred from '15.2'" }, { "name": "ytdChange", "type": [ "double", "null" ], "doc": "Type inferred from '-0.25696247383444343'" } ]}{ "type" : "record", "name" : "iextradingchart", "fields" : [ { "name" : "date", "type" : ["string","null"] }, { "name" : "open", "type" : ["double","null"] }, { "name" : "high", "type" : ["double","null"] }, { "name" : "low", "type" : ["double","null"] }, { "name" : "close", "type" : ["double","null"] }, { "name" : "volume", "type" : ["int","null"] }, { "name" : "unadjustedVolume", "type" : ["int","null"] }, { "name" : "change", "type" : ["double","null"] }, { "name" : "changePercent", "type" : ["double","null"] }, { "name" : "vwap", "type" : ["double","null"] }, { "name" : "label", "type" : ["string","null"] }, { "name" : "changeOverTime", "type" : ["int","null"] } ] } Messages to Slack File: ${'filename'} Offset: ${'kafka.offset'} Partition: ${'kafka.partition'} Topic: ${'kafka.topic'} UUID: ${'uuid'} Record Count: ${'record.count'} File Size: ${fileSize:divide(1024)}K See jsonpath.com Splits $.*.quote $.*.chart $.*.news Array to Single $.* GETHTTP URL https://api.iextrading.com/1.0/stock/market/batch?symbols=hdp&types=quote,news,chart&range=1y&last=25000 FileName marketbatch.hdp.${'hdp':append(${now():format('yyyymmddHHMMSS'):append(${md5}):append('.json')})} Data provided for free by IEX. View IEX’s Terms of Use. IEX Real-Time Price https://iextrading.com/developer/ Queries SELECT * FROM FLOWFILE WHERE latestPrice > week52Low SELECT * FROM FLOWFILE WHERE latestPrice <= week52Low Example Output File: 855957937589894 Offset: 22460 Partition: 0 Topic: iextradingquote UUID: b2a8e797-2249-4689-9a78-4339ddb5ecb4 Record Count: File Size: 3K Data Visualization in Apache Zeppelin with Hive and Spark SQL Creating tables on top of Apache ORC files in HDFS is easy. Push Some Messages to Slack Resources https://phoenix.apache.org/hive_storage_handler.html https://github.com/aol/druid/tree/master/docs/_graphics Other Data Sources https://www.kaggle.com/qks1lver/amex-nyse-nasdaq-stock-histories https://github.com/qks1lver/redtide Source https://github.com/tspannhw/stocks-nifi-kafka stocks-copy.jsonstock-to-kafka.xml ... View more

In preparation for my talk at the Philadelphia Open Source Conference(https://phillyopensource.splashthat.com/), Apache Deep Learning 201, I wanted to have some good images for running various Apache MXNet GluonCV Deep Learning Algorithms for Computer Vision. See: https://gluon-cv.mxnet.io/ Using Apache open source tools - Apache NiFi 1.8 and Apache MXNet 1.3 with GluonCV I can easily ingest live traffic camera images and run Object Detection, Semantic Segmentation and Instance Segmentation. Code: https://github.com/tspannhw/ApacheDeepLearning201 It's so easy, I am running multiple on the data to see which gives me the results I like. I am like YOLO which is a one of my old favorites. yolonifitraffic.py demo_mask_rcnn_nyc.py demo_deeplab_nyc.py So we can find the cars and people in these webcams. Use cases can be around traffic optimization, public safety and advertisement optimization. Due to licensing, I thought it better not to show Traffic Camera data here. To industrialize and scale out this process from a single Data Scientist to a national ingestion system, we use the power of Apache NiFi to ingest, process and control flows. I am using the latest Apache NiFi 1.8. Apache NiFi Flow to Ingest and Process Traffic Camera Data First we have a list of URLs that I want to process, this can be sourced and stored anywhere. For ease of use with a static set I am using GenerateFlowFile. I have a JSON file of URLs that I split and parse to call various Computer Vision Python scripts (DeepLab3, MaskRCNN, YOLO and others). YOLO seems to be the most useful so far. I am grabbing the results, some system metrics, metadata and the deep learning analytics generated by Apache MXNet. I split the flow into two portions. One builds GluonCV result data from YOLO and the other creates a file from TensorFlow results done on the fly. Here is a list of my webcam URLs. There's millions of them out there. If your data is tabular, then you need a schema for fast record processing. An Example Dataset Returned from GLUONCV - YOLO Python 3.6 Script I turn JSON data into HDFS Writeable AVRO Data and Can Run Live SQL on It One Output Source Code Be a Joint Slack Group Object Detection: GluonCV YOLO v3 and Apache NiFi This can be OpenCV, a static photo or from a URL. Object Detection: Faster RCNN with GluonCV Faster RCNN model trained on Pascal VOC dataset with ResNet-50 backbone net = gcv.model_zoo.get_model(faster_rcnn_resnet50_v1b_voc, pretrained=True) https://gluon-cv.mxnet.io/api/model_zoo.html Instance Segmentation: Mask RCNN with GluonCV Mask RCNN model trained on COCO dataset with ResNet-50 backbone net = model_zoo.get_model('mask_rcnn_resnet50_v1b_coco', pretrained=True) https://gluon-cv.mxnet.io/build/examples_instance/demo_mask_rcnn.html https://github.com/matterport/Mask_RCNN https://arxiv.org/abs/1703.06870 Photo by Ryoji Iwata on Unsplash There's a lot of people crossing the street! Semantic Segmentation: DeepLabV3 with GluonCV GluonCV DeepLabV3 model on ADE20K dataset model = gluoncv.model_zoo.get_model('deeplab_resnet101_ade', pretrained=True) run1.sh demo_deeplab_webcam.py This runs pretty slow on a machine with no GPU. https://www.cityscapes-dataset.com/ http://groups.csail.mit.edu/vision/datasets/ADE20K/ https://arxiv.org/abs/1706.05587 https://gluon-cv.mxnet.io/build/examples_segmentation/demo_deeplab.html That is the best picture of me ever! Semantic Segmentation: Fully Convolutional Networks GluonCV FCN model on PASCAL VOC dataset model = gluoncv.model_zoo.get_model(‘fcn_resnet101_voc ', pretrained=True) run1.sh demo_fcn_webcam.py https://gluon-cv.mxnet.io/build/examples_segmentation/demo_fcn.html https://people.eecs.berkeley.edu/~jonlong/long_shelhamer_fcn.pdf It found me. For NYC Dot and PennDot camera usage, you have to sign a developer agreement for a feed! See: http://www.nyc.gov/html/dot/html/home/home.shtml https://www.penndot.gov/Pages/default.aspx ... View more