Building a Custom Processor in Apache NiFi for TensorFlow Using the Java API

TensorFlow has released a Java API, so I decided to write a quick custom processor to run TensorFlow Inception v3.

It's a simple set of dependencies for Maven:

It's easy to add the new processor NiFi. First, build it using mvn install (see my build script), then deploy it:

cp nifi-tensorflow-nar/target/nifi-tensorflow-nar-1.0.nar /Volumes/Transcend/Apps/nifi-1.2.0/lib

Once you restart NiFi, you can add the TensorFlow Processor.

An example flow is to the use the very smart ListFile , which will iterate through a list of files and keep track of the timestamp of files it last accessed. I point to a directory of files and the NiFi processor gets fed a ton of images to very quickly process. This is much faster than my calling out to a script.

I gave it a picture of my RV...

And this is what it guessed. The result of the run is a new attribute, probabilities, which is a string description of what it could be a confidence percentage as text:

Here's the source code.

Resources

• Data Lake 3.0: Containerization, Erasure Coding, GPU Pooling n
• HDF 2.0 Flow for Processing Real-Time Tweets
• Using an ASUS Tinkerboard with TensorFlow and Python
• Setting Up GPU-Enabled TensorFlow to Work With Zeppelin
• IoT Capturing Photos and Analyzing the Image With TensorFlow on a Raspberry Pi
• Analyzing Images in HDF 2.0 With TensorFlow
• Deep Learning IoT Workflows With Raspberry Pi MQTT