In this repos we are trying to solve the classification task using the wide RESNET family architecture models with pytorch.
We are using the state-of-the-art dataset call INTEL IMAGE CLASSIFICATION .
Resnet article is the most cited in the deep learning domain. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth.This result won the 1st place on the ILSVRC 2015 classification task. we have used resnet50 , resnet101 , resnet152 and other main resnet family architecture for showing how these method works on this dataset for classification.
This Data contains around 25k images of size 150x150 distributed under 6 categories:
'buildings' -> 0,
'forest' -> 1,
'glacier' -> 2,
'mountain' -> 3,
'sea' -> 4,
'street' -> 5
1. Resnet (https://arxiv.org/abs/1512.03385)
2. Wide Resnet (https://arxiv.org/abs/1605.07146)
3. ResNext (https://arxiv.org/abs/1611.05431)
4. DenseNet (https://arxiv.org/abs/1608.06993)
5. Inception Resnet (https://arxiv.org/abs/1602.07261)