Research map / Detection and segmentation / CNN features
Hardware-in-the-Loop End-to-End Optimization of Camera Image Processing Pipelines
Ali Mosleh, Avinash Sharma, Emmanuel Onzon, Fahim Mannan, Nicolas Robidoux, Felix Heide
Check 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 |
|---|---|---|---|
| Optimize ISP hyperparameters end-to-end for object-detection loss rather than perceptual image quality | Synthetic ISP · COCO, eighty-class object detection | mAP: 0.35 -> 0.39 | Improved |
| Optimize ISP hyperparameters end-to-end for object-detection loss rather than perceptual image quality | Synthetic ISP · KITTI, automotive object detection | mAP: 0.61 -> 0.75 | Improved |
| Use hardware-in-the-loop optimization instead of an ISP differentiable approximation | ARM Mali-C71 ISP · KITTI, automotive object detection | mAP: 0.26 -> 0.44 | Improved |
| Optimize ISP hyperparameters for segmentation loss rather than perceptual image quality | Synthetic ISP · COCO, instance segmentation | mAP: 0.26 -> 0.32 | Improved |
| Optimize ISP hyperparameters for segmentation loss rather than perceptual image quality | Synthetic ISP · COCO, panoptic segmentation | PQ: 0.28 -> 0.35 | Improved |
| Use hardware-in-the-loop optimization instead of an ISP differentiable approximation | ARM Mali-C71 ISP · Rainbow chart | PSNR: 19.53 -> 21.52 | Improved |
| Use hardware-in-the-loop optimization instead of an ISP differentiable approximation | OnSemi AP0202AT ISP · Rainbow chart | CIELAB ΔE: 10.40 -> 9.88 | Improved |
| Jointly optimize ISP hyperparameters rather than optimizing blocks sequentially | Synthetic ISP · COCO, eighty-class object detection | mAP: 0.20 -> 0.39 | Improved |
Closest papers by meaning
- Learning To Exploit the Sequence-Specific Prior Knowledge for Image Processing Pipelines Optimization (CVPR 2023 · similarity 0.91)
- AdaptiveISP: Learning an Adaptive Image Signal Processor for Object Detection (NeurIPS 2024 · similarity 0.86)
- RL-SeqISP: Reinforcement Learning-Based Sequential Optimization for Image Signal Processing (AAAI 2024 · similarity 0.85)
- RGB-Event ISP: The Dataset and Benchmark (ICLR 2025 · similarity 0.83)
- ParamISP: Learned Forward and Inverse ISPs using Camera Parameters (CVPR 2024 · similarity 0.82)
- Learning Degradation-Independent Representations for Camera ISP Pipelines (CVPR 2024 · similarity 0.81)
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.
