[Paper] CAM: Learning Deep Features for Discriminative Localization (Weakly Supervised Object Localization)

Revisit Global Average Pooling (GAP), Weakly Supervised Object Localization While Image Classification, Outperforms Backprop

Class Activation Mapping (CAM)

Outline

1. Class Activation Mapping (CAM) Using Global Average Pooling (GAP)

1.1. Network Using GAP for CAM

Class Activation Mapping (CAM) Using Global Average Pooling (GAP)
The CAMs of two classes from ILSVRC

1.2. Weakly Supervised Object Localization (WSOL)

Examples of localization from GoogLeNet-GAP

1.3. GAP vs GMP

2. AlexNet, VGGNet, GoogLeNet Using GAP for CAM

3. Experimental Results

3.1. Classification

Classification error on the ILSVRC validation set.

3.2. Weakly-Supervised Object Localization (WSOL)

Localization error on the ILSVRC validation set.
Class activation maps from CNN-GAPs and the class-specific saliency map from the Backprop methods.
Localization error on the ILSVRC test set

3.3. Deep Features for Generic Localization

Classification accuracy on representative scene and object datasets for different deep features.

3.4. Fine-grained Recognition

Fine-grained classification performance on CUB200
CAMs and the inferred bounding boxes (in red) for selected images from four bird categories in CUB200.

Reference

Weakly Supervised Object Localization (WSOL)

My Other Previous Paper Readings

PhD, Researcher. I share what I've learnt and done. :) My LinkedIn: https://www.linkedin.com/in/sh-tsang/, My Paper Reading List: https://bit.ly/33TDhxG