Texture Classification of Leather Defects

A very basic project to binary classify image samples of leather texture into defective or non-defective.

A little about the main concept of convolutional layers used in image processing, that I have employed as a layer while training the model. The main purpose to pass the image through multiple layers is to identify the defects clearly, which are recognised as pixel deviations from its surroundings.

A broad outline of the layers used in the model and their working

The sequential model consisted of the following layers

Dropout Layer

Conv2D Layer (as explained above)

MaxPooling2D Layer

This layer is placed on the convulated image after applying the convulution layer. Picks up the maximum intensity values, hence the word 'Max'.

Flatten Layer

This was used to convert the multi-dimensional output from the above layers to one-dimensional value.

Dense Layer

This was used for classification purposes.