Deep-Learning-Animal-Classification-using-TensorFlow

Deep Learning Animal Classification using TensorFlow

Deep Learning - Beginner Challenge from HackerEarth (https://www.hackerearth.com/challenge/competitive/deep-learning-beginner-challenge/)

Identify the Animal

Problem Statement

Wildlife images captured in a field represent a challenging task while classifying animals since they appear with a different pose, cluttered background, different light and climate conditions, different viewpoints, and occlusions. Additionally, animals of different classes look similar. All these challenges necessitate an efficient algorithm for classification.

In this challenge, you will be given 19,000 images of 30 different animal species. Given the image of the animal, your task is to predict the probability for every animal class. The animal class with the highest probability means that the image belongs to that animal class.

I am using TensorFlow to do image classification here is the Blog and which i learned https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0

please follow the instruction to install tensorFlow python Extract the SRC Zip file since i am using MobileNet CNN which works on 244x244 image, so resize(scaling, re-sampling) the image to achieve 244x244. it will aspect ratio so that the images wont stretch. This code is to resize the image and save it in tensorFlow Directory arrange it in folder vise. species name is the folder name

python -m scripts.load_image --train_folder=C:/my/test/train/ --train_file=C:/my/test/train.csv

Then for training following command

for windows

set IMAGE_SIZE=224 set ARCHITECTURE=mobilenet_1.0_%IMAGE_SIZE%

For CNN iam using is Mobilenet_1.0_224

python -m scripts.retrain --bottleneck_dir=tf_files/bottlenecks --model_dir=tf_files/models/ --summaries_dir=tf_files/training_summaries/"%ARCHITECTURE%" --output_graph=tf_files/retrained_graph.pb --output_labels=tf_files/retrained_labels.txt --architecture="%ARCHITECTURE%" --how_many_training_steps 500 --image_dir=tf_files/animals

"how_many_training_steps 500" in 500 steps we will get 95% accuracy, for more accuracy you can increase steps up to 4000(max limit)

This is for testing. Following code will resize the test image to provide a proper analyses and accuracy

python -m scripts.load_test_image --test_folder=C:/my/test/test/ --test_file=C:/my/test/test.csv

Finally run this code to get the submissions. check the result.csv in meta-data folder inside tensorflow

python -m scripts.test_submission --test_file=C:/my/test/test.csv