Research map / Detection and segmentation / CNN features
A Novel Visual Interpretability for Deep Neural Networks by Optimizing Activation Maps with Perturbation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Apply SIGCAM | ImageNet classifiers · ImageNet-1k val, Inception v3 | insertion AUC: 56.14 -> 57.94 | Helped |
| Apply SIGCAM | ImageNet classifiers · ImageNet-1k val, Densenet201 and Vgg19 | insertion AUC: 58.65 -> 54.61; 53.50 -> 42.54 | Hurt |
| Apply SIGCAM | ImageNet classifiers · ImageNet-1k val, Densenet201, Inception v3, and Vgg19 | deletion AUC: 10.63 -> 10.97; 9.32 -> 10.22; 7.38 -> 7.80 | Mixed |
| Apply SIGCAM | ResNet50 classifier · COCO2017 val2017 pointing game | mean accuracy: +2.92% | Helped |
| Add WGradCAM feature-map activations as pixel-wise gradient-weight factors (ablation) | GradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | deletion AUC: 17.92 -> 17.23 | Helped |
| Add WGradCAM feature-map activations as pixel-wise gradient-weight factors (ablation) | GradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | insertion AUC: 66.04 -> 66.83 | Helped |
| Optimize a GradCAM base-mask with EMP (ablation) | GradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | over-all AUC: 48.12 -> 50.68 | Helped |
| Optimize a GradCAM base-mask with I-GOS (ablation) | GradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | over-all AUC: 48.12 -> 56.50 | Helped |
| Use WGradCAM as the base-mask and optimize it with I-GOS (ablation) | WGradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | over-all AUC: 49.60 -> 57.66 | Helped |
| Use WGradCAM as the base-mask and optimize it with I-GOS (ablation) | WGradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | insertion AUC: 66.83 -> 67.57 | Helped |
| Use WGradCAM as the base-mask and optimize it with I-GOS (ablation) | WGradCAM · ImageNet-1k, 5,000 sampled images, Inception v3 | deletion AUC: 17.23 -> 9.91 | Hurt |
Closest papers by meaning
- Rethinking Interpretation: Input-Agnostic Saliency Mapping of Deep Visual Classifiers (AAAI 2023 · similarity 0.89)
- DD-CAM: Minimal Sufficient Explanations for Vision Models Using Delta Debugging (NeurIPS 2026 · similarity 0.88)
- DiffCAM: Data-Driven Saliency Maps by Capturing Feature Differences (CVPR 2025 · similarity 0.87)
- CAPE: CAM as a Probabilistic Ensemble for Enhanced DNN Interpretation (CVPR 2024 · similarity 0.87)
- Show and Tell: Visually Explainable Deep Neural Nets via Spatially-Aware Concept Bottleneck Models (CVPR 2025 · similarity 0.86)
- Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.