Research map / Video / Action recognition
Can an Image Classifier Suffice For Action Recognition?
Quanfu Fan, Chun-Fu Chen, Rameswar Panda
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Use a square 3x3 super-image layout instead of a 1x8 strip (ablation) | SIFAR-B-12 · SSV2 | top-1 accuracy: 44.4 -> 60.8 | Improved |
| Use the 3x3 grid rather than the 2x4 layout in Fig. b (ablation) | SIFAR-B-12 · SSV2 | top-1 accuracy: 58.6 -> 60.8 | Improved |
| Add absolute position embedding per frame (ablation) | SIFAR-B-12 · SSV2 | top-1 accuracy: 59.5 -> 60.8 | Improved |
| Add absolute position embedding per frame (ablation) | SIFAR-B-12 · Kinetics400 | top-1 accuracy: 80.0 -> 79.7 | Hurt |
| Reverse the temporal order of input frames (ablation) | SIFAR-B-12 · SSV2 | top-1 accuracy: 60.8 -> 23.9 | Hurt |
| Randomize the temporal order of input frames (ablation) | SIFAR-B-12 · SSV2 | top-1 accuracy: 60.8 -> 39.4 | Hurt |
| Reverse the temporal order of input frames (ablation) | SIFAR-B-12 · Kinetics400 | top-1 accuracy: 80.0 -> 79.8 | No clear effect |
| Add a 21x21-kernel block to ResNet50 (ablation) | CNN-based SIFAR · SSV2 | top-1 accuracy: 50.8 -> 55.8 (+5.0) | Improved |
| Add a 21x21-kernel block and an 11x11-kernel block to ResNet50 (ablation) | CNN-based SIFAR · SSV2 | top-1 accuracy: 50.8 -> 57.6 (+6.8) | Improved |
| Increase the CNN-based SIFAR input from 8 to 16 frames (ablation) | SIFAR-R50 · SSV2 | top-1 accuracy: 50.8 -> 49.0 | Hurt |
Closest papers by meaning
- Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition (ECCV 2016 · similarity 0.87)
- ViViT: A Video Vision Transformer (ICCV 2021 · similarity 0.87)
- Deep Analysis of CNN-Based Spatio-Temporal Representations for Action Recognition (CVPR 2021 · similarity 0.87)
- Revisiting Classifier: Transferring Vision-Language Models for Video Recognition (AAAI 2023 · similarity 0.87)
- Action Recognition With Spatial-Temporal Discriminative Filter Banks (ICCV 2019 · similarity 0.87)
- Understanding Video Transformers via Universal Concept Discovery (CVPR 2024 · 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.
