Rapidos Image classification with Tensorflow
Felt so bored after work and too lazy to do yoga, I decided to do some image classification with TensorFlow for yoga positions.
It is just some dirty testing of the new tool make_image_classifier, It is meant to replace image_retraining/retrain.py for people who already did this with the retrain script
What is Tensorflow ?
Tensorflow is the craziest deep learning library in the world right now, that made the life of many developers easy and even let some DevOps like me playing with it, it focuses on simplicity and ease of use. ( Okay! Okay ! not that easy for something useful )
We have some steps to do
- Gathering Data
I download images from google, we need as many as we can of images so we could have a good result
You can use the chrome extension Fatken, I have this as a result
2. Run Tensorflow image docker
To avoid installing python, TensorFlow, and all their dependencies since we are just testing and nothing serious, i download the TensorFlow Docker image
Since we have the dataset on our machine, we will mount it from host to docker image
3. Train the data
4. Test the data
That is all Folks!
The next episode will be about developing a Flask API to recognize clothes using Tensorflow
