Research map / Detection and segmentation / Open-set and few-shot detection
Precise Detection in Densely Packed Scenes
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 EM-Merger to the base detector using Soft-IoU scores (ablation) | ResNet-50 FPN detector · SKU-110K test | AP: .413 -> .492 | Helped |
| Add Soft-IoU scores to EM-Merger instead of using objectness scores (ablation) | ResNet-50 FPN with EM-Merger · SKU-110K test | AP: .482 -> .492 | Helped |
| Add Soft-IoU prediction while retaining standard NMS (ablation) | ResNet-50 FPN with NMS · SKU-110K test | AP: .413 -> .418 | No clear effect |
| Replace standard NMS with EM-Merger using Soft-IoU scores (ablation) | Soft-IoU detector · SKU-110K test | AP^.75: .386 -> .556 | Helped |
| Add Soft-IoU scores to EM-Merger instead of using objectness scores (ablation) | ResNet-50 FPN with EM-Merger · SKU-110K test | MAE: 23.978 -> 14.522 | Helped |
| Replace standard NMS with EM-Merger using Soft-IoU scores (ablation) | Soft-IoU detector · SKU-110K test | MAE: 25.394 -> 14.522 | Helped |
| Use the full approach rather than RetinaNet | Object detection and counting · SKU-110K test | AP: .455 -> .492 | Helped |
| Use the full approach rather than RetinaNet | Object counting · CARPK | MAE: 16.62 -> 6.77 | Helped |
| Use the full approach rather than IEP Counting | Object counting · PUCPR+ | MAE: 15.17 -> 7.16 | Helped |
| Add Soft-IoU layer (ablation) | Base detector · SKU-110K | FPS: 0.5 -> 0.5 | No clear effect |
| Add CPU EM-Merger (ablation) | Base detector · SKU-110K | FPS: 0.5 -> 0.23 | Hurt |
Closest papers by meaning
- Bottom-Up Object Detection by Grouping Extreme and Center Points (CVPR 2019 · similarity 0.87)
- Don't Even Look Once: Synthesizing Features for Zero-Shot Detection (CVPR 2020 · similarity 0.87)
- VarifocalNet: An IoU-Aware Dense Object Detector (CVPR 2021 · similarity 0.86)
- Unsupervised Object Detection with Theoretical Guarantees (NeurIPS 2024 · similarity 0.85)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.85)
- Cubify Anything: Scaling Indoor 3D Object Detection (CVPR 2025 · similarity 0.85)
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.