Research map / Detection and segmentation / CNN features
Part-Stacked CNN for Fine-Grained Visual Categorization
Check 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 |
|---|---|---|---|
| Add a 1×1 conv6 layer after conv5 (ablation) | PS-CNN localization network · CUB-200-2011 test | APK: 83.5 -> 84.7 | Helped |
| Increase conv6 output channels from 256 to 512 (ablation) | PS-CNN localization network · CUB-200-2011 test | APK: 84.7 -> 84.8 | Helped |
| Add Gaussian smoothing to the localization heatmaps (ablation) | PS-CNN localization network · CUB-200-2011 test | MPK: 71.5 -> 80.0 | Helped |
| Add 2 object parts instead of using bounding-box supervision only (ablation) | PS-CNN · CUB-200-2011 test | accuracy: 69.08 -> 73.72 | Helped |
| Increase the number of object parts from 2 to 4 (ablation) | PS-CNN · CUB-200-2011 test | accuracy: 73.72 -> 74.84 | Helped |
| Increase the number of object parts from 4 to 8 (ablation) | PS-CNN · CUB-200-2011 test | accuracy: 74.84 -> 76.22 | Helped |
| Increase the number of object parts from 8 to 15 (ablation) | PS-CNN · CUB-200-2011 test | accuracy: 76.22 -> 76.15 | No clear effect |
| Initialize the added convolution using PCA rather than the auxiliary-CNN initialization (ablation) | PS-CNN classification network · CUB-200-2011 | accuracy | Hurt |
Closest papers by meaning
- Interpretable and Accurate Fine-grained Recognition via Region Grouping (CVPR 2020 · similarity 0.92)
- Fine-Grained Image Classification via Combining Vision and Language (CVPR 2017 · similarity 0.91)
- SPDA-CNN: Unifying Semantic Part Detection and Abstraction for Fine-Grained Recognition (CVPR 2016 · similarity 0.91)
- Fine-Grained Recognition as HSnet Search for Informative Image Parts (CVPR 2017 · similarity 0.90)
- Weakly Supervised Complementary Parts Models for Fine-Grained Image Classification From the Bottom Up (CVPR 2019 · similarity 0.89)
- Democratizing Fine-grained Visual Recognition with Large Language Models (ICLR 2024 · similarity 0.88)
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.