README.md

-bitbucket

#image blur detection

i am putting it in the public format

here features are image size and number of pixel it occupies and shape

#the project consist four parts #1 classsification-KNN classifier #2 traing #3 testing and validation #4 find accuracy #here we use train and test the data in dataset #first convert the data folder into .jar file #install openCV and then create file name as detect_blur.py #import packages tensorflow,sklearn,keras and imulits and gensim #classification is run as $python ml6-img_classifier.py #execuete code as python detect_blur.py

it can be trained by tensorflow

#for that one we have to download tensorflow package #then train run script as #$ python -m scripts.retrain
--bottleneck_dir=tf_files/bottlenecks
--how_many_training_steps=500
--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}"
--image_dir=tf_files/flower_photos

now for find the accuracy

#$ python detect_blur.py