Research map / Detection and segmentation / CNN features
SCNN: A General Distribution Based Statistical Convolutional Neural Network with Application to Video Object Detection
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 |
|---|---|---|---|
| Replace VGG16 backbone with SCNN (m=8, N=16) | VGG16 video object detector · DAC 2018 video object detection | FPS: 178% speedup | Helped |
| Replace VGG16 backbone with SCNN (m=8, N=16) | VGG16 video object detector · DAC 2018 video object detection | mAP: 4.8% drop | Hurt |
| Use SCNN with m=8 instead of deterministic CNN | VGG11 video object detector · DAC 2018 video object detection | mAP: 62.2 -> 56.5 | Hurt |
| Use SCNN with m=8 instead of deterministic CNN | VGG11 video object detector · DAC 2018 video object detection | FPS: 101 -> 181 | Helped |
| Use SCNN with m=8 instead of deterministic CNN | ResNet18 video object detector · DAC 2018 video object detection | mAP: 65.1 -> 57.4 | Hurt |
| Use SCNN with m=8 instead of deterministic CNN | ResNet18 video object detector · DAC 2018 video object detection | FPS: 247 -> 363 | Helped |
| Increase basis dimension from m=8 to m=14 (ablation) | SCNN video object detector · DAC 2018 video object detection | mAP: 56.5 -> 58.7; 57.1 -> 58.3; 57.4 -> 61.9; 56.7 -> 62.2 | Helped |
| Increase basis dimension from m=8 to m=14 (ablation) | SCNN video object detector · DAC 2018 video object detection | FPS: 181 -> 116; 116 -> 72; 363 -> 276; 246 -> 188 | Hurt |
| Use SCNN instead of CNN | VGG16 video object detector · car category | mAP: 60.7 -> 76.6 | Helped |
| Use SCNN instead of CNN | VGG16 video object detector · riding category | mAP: 80.4 -> 86.9 | Helped |
| Use SCNN instead of CNN | VGG16 video object detector · building category | mAP: 91.3 -> 73.8 | Hurt |
| Use SCNN instead of CNN | VGG16 video object detector · horseride category | mAP: 40.5 -> 24.3 | Hurt |
Closest papers by meaning
- R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016 · similarity 0.90)
- Contextual Convolutional Networks (ICLR 2023 · similarity 0.90)
- Sparse R-CNN: End-to-End Object Detection With Learnable Proposals (CVPR 2021 · similarity 0.89)
- CNN Filter DB: An Empirical Investigation of Trained Convolutional Filters (CVPR 2022 · similarity 0.89)
- TGBFormer: Transformer-GraphFormer Blender Network for Video Object Detection (AAAI 2025 · similarity 0.88)
- AsCAN: Asymmetric Convolution-Attention Networks for Efficient Recognition and Generation (NeurIPS 2024 · similarity 0.87)
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.