HandPose_Detection
Background Subtraction for pose detection using Convolutional Neural Network The work is inspired from https://gogul09.github.io/software/hand-gesture-recognition-p1
Dataset Preparation
Run python2 dataset.py to prepare the dataset
Classes=Super,Palm,Fist,Thumbs_up,Victory,Why
run the code after 30 frame pose your hand in the bounding box displayed on the video
Training
Then Resize the images using resize.py
The file read.py helps to find the image resolution and no of channels which can be helpful reshaping
set some default parameters
Filters=32
EPOCHS=50
Activation_fn='relu'
Image_width=100
Image_height=100
Optimizer='adam'
Learning_rate=0.001
Snapshot=100
Num_classes=6
stride=2
Then Run python2 train.py to train the dataset
Results
The results were uploaded in the above
Run predict.py to detect