Hand Gesture Controlled Robot Using OpenCV

Vision-based gesture recognition is an important means of human-computer interaction. This paper designs a gesture based robot system using a web-camera. After obtaining the image through the camera, various image processing techniques are performed on the gesture to remove noises, which include other operations such as binarization, erosion, dilation smoothing etc.., Then the processed gesture is given as an input to CNN model for Recognition.VGG16 architecture is used for the training CNN model.VGG16 model is loaded from keras and it is trained with our dataset and validation of model is also done. After the creation of model the input gesture captured using a web-camera which is also subjected to processing mentioned earlier. Then it is given as an input to the CNN model. The model will recognize the gesture, this recognized gesture is serially transferred to arduino board. Arduino board which in turn control the bot movement.