Ablation-CAM: Visual Explanations for Deep Convolutional Network via Gradient-free Localization
| Autor: | Saurabh Desai, Harish G. Ramaswamy |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Contextual image classification
Computer science business.industry Feature extraction Pattern recognition 02 engineering and technology Image segmentation 010501 environmental sciences 01 natural sciences Convolutional neural network Backpropagation Visualization Discriminative model Feature (computer vision) 020204 information systems 0202 electrical engineering electronic engineering information engineering Artificial intelligence business 0105 earth and related environmental sciences |
| Zdroj: | WACV |
| DOI: | 10.1109/wacv45572.2020.9093360 |
| Popis: | In response to recent criticism of gradient-based visualization techniques, we propose a new methodology to generate visual explanations for deep Convolutional Neural Networks (CNN) - based models. Our approach – Ablation-based Class Activation Mapping (Ablation CAM) uses ablation analysis to determine the importance (weights) of individual feature map units w.r.t. class. Further, this is used to produce a coarse localization map highlighting the important regions in the image for predicting the concept. Our objective and subjective evaluations show that this gradient-free approach works better than state-of-the-art Grad-CAM technique. Moreover, further experiments are carried out to show that Ablation-CAM is class discriminative as well as can be used to evaluate trust in a model. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|