Android app containing an Image classifier based on transfer learning CNN using Tensorflow 1.4.1 on Stanford's Imagenet cars dataset.
- This application is a demonstration of an image classifier built using convolutional neural network.
- The model is trained on Stanford's imagenet dataset of 196 cars. Dataset can be accessed here: http://ai.stanford.edu/~jkrause/cars/car_dataset.html
- The Cars dataset contains 16,185 images of 196 classes of cars. Full list of cars is present here: https://paste.ubuntu.com/26311458/
- This project is published as an Android app available on Play Store. LINK
- For further details, please have a look at the Medium blogpost I wrote for this project. https://medium.com/@sumit.arora/training-a-neural-network-using-mobilenets-in-tensorflow-for-image-classification-on-android-14f2792f64c1
IMAGE_SIZE=224 ARCHITECTURE="mobilenet_0.75_${IMAGE_SIZE}"
python -m scripts.retrain
--bottleneck_dir=tf_files/bottlenecks
--how_many_training_steps=5000
--model_dir=tf_files/models/"${ARCHITECTURE}"
--summaries_dir=tf_files/training_summaries/"${ARCHITECTURE}"
--output_graph=tf_files/retrained_graph.pb
--output_labels=tf_files/retrained_labels.txt
--architecture="${ARCHITECTURE}"
--image_dir=tf_files/dataset
python -m scripts.retrain
--image_dir=tf_files/dataset
--learning_rate=0.0001
--testing_percentage=20
--validation_percentage=20
--train_batch_size=32
--validation_batch_size=-1
--flip_left_right True
--random_scale=30
--random_brightness=30
--eval_step_interval=100
--how_many_training_steps=600
--architecture mobilenet_1.0_224