Research map / Video / Video segmentation and tracking
Free Supervision From Video Games
Philipp Krähenbühl
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 albedo images instead of original images as input | FlowNet 2 · GTA V mean | flow accuracy at 5px (%): 48.86 -> 51.30 | Improved |
| Use albedo images instead of original images as input | FlowFields · GTA V mean | flow accuracy at 5px (%): 37.40 -> 44.46 | Improved |
| Use KITTI-trained depth models on GTA V | Eigen et al. and DiW · GTA V | RMSE: Mean depth 7.12; Eigen et al. 7.49; DiW 7.09 | Hurt |
| Tune reflectance-filtering parameters for GTA V | Reflectance filtering · GTA V | LMSE: 0.086 -> 0.076 | Improved |
| Combine the original image with an additional visual modality | GTA V imitation-learning agent · GTA V car action prediction | classification error (%) | Improved |
| Combine all available modalities | GTA V imitation-learning agent · GTA V car action prediction | classification error (%) | Improved |
| Use optical flow alone as input | GTA V imitation-learning agent · GTA V car action prediction | classification error (%) | Hurt |
Closest papers by meaning
- Playing for Data: Ground Truth from Computer Games (ECCV 2016 · similarity 0.91)
- Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play (ICLR 2026 · similarity 0.85)
- Video2Game: Real-time Interactive Realistic and Browser-Compatible Environment from a Single Video (CVPR 2024 · similarity 0.84)
- GameVerse: Can Vision-Language Models Learn from Video-based Reflection? (ICML 2026 · similarity 0.84)
- Diffusion Models Are Real-Time Game Engines (ICLR 2025 · similarity 0.84)
- Depth Any Video with Scalable Synthetic Data (ICLR 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.
