TensorPy

Easy Image Classification with TensorFlow

(Watch the 2-minute tutorial on YouTube)

Now runs much faster since video released!

You can use TensorPy to classify images by simply passing a URL on the command line, or by using TensorPy in your Python programs. TensorFlow does all the image-recognition work. TensorPy also simplifies TensorFlow installation by automating several setup steps into a single script (See install.sh for details).

(Read how_tensorpy_works for a detailed explanation of how TensorPy works.)

Setup Steps for Mac & Ubuntu/Linux

(Windows & Docker users: See the Docker ReadMe for running on a Docker machine. Windows requires Docker to run TensorFlow.)

Step 1: Create and activate a virtual environment named "tensorpy"

If you're not sure how to create a virtual environment, follow these instructions to learn how.

Step 2: Clone the TensorPy repo from GitHub

git clone https://github.com/TensorPy/TensorPy.git
cd TensorPy

Step 3: Install TensorPy, TensorFlow, and ImageNet/Inception

The install.sh script installs everything you need:

./install.sh

Step 4: Run the examples

Classify a single image from a URL:

classify "http://cdn2.hubspot.net/hubfs/100006/happy_animal.jpg"

Classify all images on a web page:

classify "https://github.com/TensorPy/TensorPy/tree/master/examples/images"

Classify a single image URL from a Python script: (See test_python_classify.py for details.)

cd examples
python test_python_classify.py

---

Future Work:

Eventually, the headline will change from "Image Classification with TensorFlow made easy!" to "Machine Learning with TensorFlow made easy!" once I expand on TensorPy to make other features of TensorFlow easier too. Stay tuned for updates!