Research map / Learning theory and algorithms / Prediction and decision losses
Analyzing Computer Vision Data - The Good, the Bad and the Ugly
Check 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 |
|---|---|---|---|
| evaluate only hazard frames instead of random frames | stereo algorithm evaluation · Sintel, 13 frames | algorithm distinction | No clear effect |
| select hazard frames rather than random picking | test data selection · Sintel | diagnostic insight | Helped |
| compare hazard frames with non-hazard frames by difficulty | 8 stereo algorithms · HCI, KITTI, Middlebury, Sintel, Freiburg | fraction in difficulty bins D/E | Helped |
| easiest-first frame subset | cumulative average performance · Sintel | bad pixel metric | Hurt |
| restrict hazard review to expert-agreed entries | hazard list · 8 dataset groups | hazard entries uncovered: 76 hazards, 48 unique, 50 uncovered | Mixed |
| use two analysts per dataset | HAZOP analysis · stereo datasets | bias | Helped |
| focus on hardest frames | evaluation · stereo | metric bias | Hurt |
Closest papers by meaning
- DrivingStereo: A Large-Scale Dataset for Stereo Matching in Autonomous Driving Scenarios (CVPR 2019 · similarity 0.87)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.86)
- What Makes Good Synthetic Training Data for Zero-Shot Stereo Matching? (CVPR 2026 · similarity 0.86)
- RUBIK: A Structured Benchmark for Image Matching across Geometric Challenges (CVPR 2025 · similarity 0.84)
- OpenAD: Open-World Autonomous Driving Benchmark for 3D Object Detection (NeurIPS 2025 · similarity 0.84)
- BEHAVIOR Vision Suite: Customizable Dataset Generation via Simulation (CVPR 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.