Research map / Learning theory and algorithms / Combinatorial and search optimization
DISeR: Designing Imaging Systems with Reinforcement Learning
Tzofi Klinghoffer, Kushagra Tiwary, Nikhil Behari, Bhavya Agrawalla, Ramesh Raskar
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 |
|---|---|---|---|
| use RL-designed camera rig instead of nuScenes rig (single camera) | CVT BEV segmentation in CARLA · CARLA Expt. a | IoU: 0.267 -> 0.341 | Improved |
| use RL-designed camera rig instead of nuScenes rig (up to six cameras) | CVT BEV segmentation in CARLA · CARLA Expt. b | IoU: 0.355 -> 0.427 | Improved |
| RL search instead of random search for camera rigs (ablation) | camera designer · CARLA BEV segmentation | IoU: 0.084 -> 0.427 (random rig vs ours, Expt. b) | Improved |
| jointly train perception model with camera designer instead of freezing the pre-trained PM (ablation) | DISeR · CARLA BEV segmentation | IoU reward | Improved |
| increase coverage (cameras viewing the object) from 0 to 3 (ablation) | stereo depth camera designer · Stereo depth of sphere | L1 loss: 14.0 -> 5.7 | Improved |
| use multiple cameras instead of one in monocular-cue-free stereo (ablation) | perception model · Stereo depth of sphere | L1 loss | Improved |
| add penalty per added camera and forward-only vehicles (ablation) | camera designer · CARLA Expt. c | number of cameras | Improved |
| move the two rear-placed front-facing cameras to the front (ablation) | Expt. b rig · CARLA Expt. b | IoU: IoU 5% better when cameras are in the back | Hurt |
Closest papers by meaning
- PeRL: Permutation-Enhanced Reinforcement Learning for Interleaved Vision-Language Reasoning (NeurIPS 2025 · similarity 0.85)
- LookThere! Sparse Vision by Reinforced Selection (NeurIPS 2026 · similarity 0.85)
- SenseSearch: Empowering Vision-Language Models with High-Resolution Agentic Search-Reasoning via Reinforcement Learning (CVPR 2026 · similarity 0.84)
- COVR: Collaborative Optimization of VLMs and RL Agent for Visual-Based Control (AAAI 2026 · similarity 0.84)
- MaskSense: Confronting the Visual Exploration Trap in Masked Image Generation (NeurIPS 2026 · similarity 0.84)
- RL-SeqISP: Reinforcement Learning-Based Sequential Optimization for Image Signal Processing (AAAI 2024 · 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.
