Biryani-or-Not-Biryani

Biryani or Not Biryani is a Biryani classifier inspired by Silicon Valley TV series. It predict a image is Biryani or not.

Getting Started

This instruction is for GPU less windows 10 machine. If you are using linux or mac this is more easier than that.

Prerequisites

• Anaconda
• Python 3.5
• Tensorflow

Installing

First you have to create a anaconda enviorment and activate it.

(C:\Anaconda3) C:\> conda create -n biryani python 3.5.2
(C:\Anaconda3) C:\> activate biryani

If you are using machine without GPU like me install CPU-only version of Tensorflow

(biryani)C:> pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.1.0-cp35-cp35m-win_amd64.whl 

Now it's time to train Biryani model

(biryani)C:> python retrain.py --bottleneck_dir=bottlenecks --model_dir=inception \
--summaries_dir=training_summaries/long --output_graph=retrained_graph.pb \
--output_labels=retrained_labels.txt --image_dir=images

That is gonna take a while depending on your machine. After it is done it is already download google inception model and create 3 new folder training_summaries, inception, bottlenecks and save new output graph in retrained_graph.pb and output labels in retrained_labels.txt

Running the tests

Now we have our Biryani classifier. To test this classifier run label_image.py with image dir/name

(biryani)C:> python label_image.py "Test Image\pic_006.jpg"

this will provide output something like this

biryani (score = 0.99412)
notbiryani (score = 0.00588)

The Test Image folder have some image I used for test. You can use that or use other images for your test.

Acknowledgments

• Tensorflow official tutorial
• Fatkun Batch Download Image for helping collect Train and Test image from google image search.
• And of course Silicon Valley for spread this funny idea.