Skip to content

this project is built of some customized layers to be trained on image detection tasks based on the idea of Inception

Notifications You must be signed in to change notification settings

abbass-zain-eddine/ImageClassification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ImageClassification

This project is built of some customized layers block to be trained on image classification tasks based on the idea of Inception

Customized Layers

This folder contains three customized layers and one miniInception model. The first block is the ConvModule. In this block there are two consecutive layers, the conv2D layer followed by batch normalization. The second block is the InceptionModule, which is composed of three layers. A Two ConvModule block with 1x1 and 3x3 kernels respectively, followed by a concatenation layer that concatenates the output of those two ConvModule blocks. Finally the DownsampleModule that is also composed of three parts. A ConvModule of 3x3 kernel and 2x2 strides and a 3x3 maxpooling layer also 2x2 strided. The third component is a concatenation layer that combines the output of the previous two components. There is also a MiniInceptionModule that is built up of the previous defined modules.

Train

The train folder contains a train class that is responsible for managing all the training processes of the miniInceptionModule. all you have to do is to load your data and send it to this model with some additional attributes.

main

The main.py file tests all the previous code on the cifar-10 data set.

About

this project is built of some customized layers to be trained on image detection tasks based on the idea of Inception

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages