Sign languages digits recognition with CNN

A simple Convolutional Neural Network (CNN) model is trained and used to infer the correct digit from a set of images that encode the decimal numbers in the range 0-9 in hand sign language.

The dataset used for training/inference is available here. The CNN architecture used is the SmallVGGNet, a smaller version of the VGGNet.

Sample images for each digit

0	1	2	3	4
5	6	7	8	9

Training and validation metrics