Research map / Detection and segmentation / CNN features
TorchAdapt: Towards Light-Agnostic Real-Time Visual Perception
Khurram Azeem Hashmi, Karthik Palyakere Suresh, Didier Stricker, Muhammad Zeshan Afzal
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 Torch module to the baseline (ablation) | YOLOv3 · ExDark object detection (low-light) | mAP50: 78.8 -> 79.1 | Improved |
| add Torch module to the baseline (ablation) | RTMDet-Ins-tiny · LIS instance segmentation (low-light) | SegAP: 46.0 -> 51.1 | Improved |
| add Adapt module on top of Torch (ablation) | YOLOv3 · COCO+ExDark object detection (light-agnostic) | mAP50: 62.9 -> 63.5 | Improved |
| add Adapt module on top of Torch (ablation) | RTMDet-Ins-tiny · LIS instance segmentation (low-light) | SegAP: 51.1 -> 53.3 | Improved |
| add illumination scaling without Adapt module (ablation) | YOLOv3 · COCO+ExDark object detection (light-agnostic) | mAP50: 62.9 -> 62.7 | Hurt |
| add illumination scaling without Adapt module (ablation) | RTMDet-Ins-tiny · COCO+LIS instance segmentation (light-agnostic) | SegAP: 34.8 -> 34.1 | Hurt |
| add illumination scaling to Torch + Adapt (ablation) | YOLOv3 · COCO+ExDark object detection (light-agnostic) | mAP50: 63.5 -> 64.0 | Improved |
| add illumination scaling to Torch + Adapt (ablation) | RTMDet-Ins-tiny · COCO+LIS instance segmentation (light-agnostic) | SegAP: 36.2 -> 36.6 | Improved |
| raise maximum scale factor kappa from 3 to 5 (ablation) | TorchAdapt · COCO+ExDark object detection (light-agnostic) | mAP50: 64.0 -> 61.1 | Hurt |
| use gamma curve instead of ISP darkening to synthesize low-light view (ablation) | TorchAdapt · COCO+ExDark object detection (light-agnostic) | mAP50: 61.7 -> 64.0 | Improved |
| apply off-the-shelf LLIE (Zero-DCE) as preprocessing instead of none | YOLOv3 · COCO+ExDark object detection (light-agnostic) | mAP50: 62.4 -> 57.6 | Hurt |
| train Zero-DCE end-to-end on the task instead of off-the-shelf | YOLOv3 · COCO+ExDark object detection (light-agnostic) | mAP50: 57.6 -> 62.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FeatEnHancer — FeatEnHancer: Enhancing Hierarchical Features for Object Detection and Beyond Under Low-Light Vision (ICCV 2023) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning Adaptive Lighting via Channel-Aware Guidance (ICML 2025 · similarity 0.86)
- Boosting Object Detection with Zero-Shot Day-Night Domain Adaptation (CVPR 2024 · similarity 0.85)
- Improving Visual and Downstream Performance of Low-Light Enhancer with Vision Foundation Models Collaboration (CVPR 2025 · similarity 0.85)
- Dispel Darkness for Better Fusion: A Controllable Visual Enhancer based on Cross-modal Conditional Adversarial Learning (CVPR 2024 · similarity 0.84)
- Zero-Reference Low-Light Enhancement via Physical Quadruple Priors (CVPR 2024 · similarity 0.84)
- Bright-NeRF: Brightening Neural Radiance Field with Color Restoration from Low-Light RAW Images (AAAI 2025 · similarity 0.84)
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.
