Research map / Detection and segmentation / CNN features
Integrated perception with recurrent multi-task neural networks
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 |
|---|---|---|---|
| Recurrently integrate task outputs instead of using ordinary shared-feature multi-task prediction | VGG-M Multinet · PASCAL VOC 2010 val | classification AP: 76.2 -> 77.4 | Helped |
| Recurrently integrate task outputs instead of using ordinary shared-feature multi-task prediction | VGG-M Multinet · PASCAL VOC 2010 val | object detection AP@50 IoU: 57.1 -> 57.5 | Helped |
| Recurrently integrate task outputs instead of using ordinary shared-feature multi-task prediction | VGG-M Multinet · PASCAL VOC 2010 val | part detection AP@40 IoU: 37.2 -> 38.8 | Helped |
| Recurrently integrate task outputs instead of using ordinary shared-feature multi-task prediction | VGG-M Multinet · PASCAL VOC 2007 test | classification AP: 78.9 -> 79.8 | Helped |
| Recurrently integrate task outputs instead of using ordinary shared-feature multi-task prediction | VGG-M Multinet · PASCAL VOC 2007 test | object detection AP: 60.4 -> 61.3 | Helped |
| Constrain the shared representation to 512 dimensions (ablation) | VGG-M Multinet · PASCAL VOC 2010 val | classification AP: 77.4 -> 76.8 | Hurt |
| Constrain the shared representation to 512 dimensions (ablation) | VGG-M Multinet · PASCAL VOC 2010 val | object detection AP@50 IoU: 57.5 -> 57.3 | Hurt |
| Constrain the shared representation to 512 dimensions (ablation) | VGG-M Multinet · PASCAL VOC 2010 val | part detection AP@40 IoU: 38.8 -> 38.5 | Hurt |
| Share convolutional features across tasks rather than train independent networks | VGG-M multi-task model · PASCAL VOC 2010 val | classification AP: 76.4 -> 76.2 | No clear effect |
| Share convolutional features across tasks rather than train independent networks | VGG-M multi-task model · PASCAL VOC 2010 val | object detection AP@50 IoU: 55.5 -> 57.1 | Helped |
| Feed ground-truth classification labels back during the first iteration | VGG-M Multinet · PASCAL VOC 2010 val | classification AP: 90.1 | Helped |
| Feed ground-truth classification labels back during the first iteration | VGG-M Multinet · PASCAL VOC 2010 val | object detection AP@50 IoU: 58.9 | Helped |
Closest papers by meaning
- End-To-End Multi-Task Learning With Attention (CVPR 2019 · similarity 0.88)
- A Unified Sequence Interface for Vision Tasks (NeurIPS 2022 · similarity 0.87)
- Cross-Stitch Networks for Multi-Task Learning (CVPR 2016 · similarity 0.86)
- Towards flexible perception with visual memory (ICML 2025 · similarity 0.86)
- Provable Multi-Task Representation Learning by Two-Layer ReLU Neural Networks (ICML 2024 · similarity 0.86)
- UniRepLKNet: A Universal Perception Large-Kernel ConvNet for Audio Video Point Cloud Time-Series and Image Recognition (CVPR 2024 · 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.