Part 5.4: Implementing RotationNet approach for Custom Data
#SelfSupervisedLearning #UnsupervisedLearning #RepresentationLearning #RotNet #RotNetImplementation
In his video we continue the rotnet implementation and discuss about evaluating the supervised and self-supervised model.
We can evaluate a self supervised model by
1. Transfer Learning
2. Linear Evaluation - using logistic regression for doing classification based on the features from convnet.
Part-1 the code structure and problem description: • Part 5.1: Implementing RotationNet ap...
Part-2 Preparing the data loader: • Part 5.2: Implementing RotationNet ap...
part-3 Training the model: • Part 5.3: Implementing RotationNet ap...
I am using flowers dataset from Kaggle and trying to emulate the scenario when you have some unlabeled data and a small amount of labeled data.
Paper - https://arxiv.org/abs/1803.07728
official Github - https://github.com/gidariss/FeatureLe...
My Github - https://github.com/anujshah1003/self_...
Video explaining the paper - • Part-4 Understanding Rotation Net app...
Kaggle's Flowers Dataset - https://www.kaggle.com/alxmamaev/flow...
