Research map / Detection and segmentation / Open-set and few-shot detection
Joint Learning of Object and Action Detectors
Vicky Kalogeiton, Philippe Weinzaepfel, Vittorio Ferrari, Cordelia Schmid
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| add flow stream to RGB with late score fusion (ablation) | RGB detector · A2D object detection | mAP: 63.1 -> 61.6 | Hurt |
| add flow stream to RGB with late score fusion (ablation) | RGB detector · ILSVRC VID object detection | mAP: 45.2 -> 33.9 | Hurt |
| use proposed RoI and feature-level fusion instead of late score fusion (ablation) | two-stream detector · A2D object detection | mAP: 61.6 -> 65.3 | Improved |
| use proposed fusion instead of late score fusion (ablation) | two-stream detector · ILSVRC VID object detection | mAP: 33.9 -> 48.1 | Improved |
| train multitask on objects and actions instead of objects only | detector · A2D objects | mAP: 65.3 -> 68.3 | Improved |
| train multitask on objects and actions instead of actions only | detector · A2D actions | mAP: 56.2 -> 60.0 | Improved |
| joint multitask training instead of separate detectors with multiplied scores | detector · A2D objects+actions | mAP: 43.1 -> 48.9 | Improved |
| use Cartesian product labels instead of multitask (ablation) | detector · A2D objects+actions | mAP: 48.9 -> 49.2 | No clear effect |
| use hierarchical labels instead of multitask (ablation) | detector · A2D objects+actions | mAP: 48.9 -> 49.6 | No clear effect |
| vary action loss weight over 0.5, 1, 2 (ablation) | multitask detector · A2D object-action | mAP: < 0.5% variation | No clear effect |
| use SharpMask proposals instead of GBH (ablation) | segmentation from detections · A2D object segmentation | mIoU: 42.7 -> 49.5 | Improved |
| use multitask architecture instead of separate networks with multiplied scores | relationship detector · VRD phrase detection | R@100: 11.9 -> 18.3 | Improved |
Closest papers by meaning
- End-to-End Joint Semantic Segmentation of Actors and Actions in Video (ECCV 2018 · similarity 0.87)
- Multi-Task Learning of Object States and State-Modifying Actions From Web Videos (TPAMI 2024 · similarity 0.87)
- Multi-Dataset, Multitask Learning of Egocentric Vision Tasks (TPAMI 2021 · similarity 0.86)
- Learn How to See: Collaborative Embodied Learning for Object Detection and Camera Adjusting (AAAI 2024 · similarity 0.85)
- Dual DETRs for Multi-Label Temporal Action Detection (CVPR 2024 · similarity 0.85)
- Visual Diversity and Region-aware Prompt Learning for Zero-shot HOI Detection (NeurIPS 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.
