Reproduce “Intra-frame Object Tracking by Deblatting” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jan Kotera, Denys Rozumnyi, Filip Šroubek, Jiří Matas
The "Tracking by Deblatting" (TbD) method enables accurate, high-resolution tracking of fast-moving, motion-blurred objects by recovering continuous intra-frame trajectories from motion blur. It achieved average Trajectory-IoU up to 0.701 and Recall over 0.90 on a new FMO dataset, substantially surpassing prior methods.
Key insights
- Motion blur, often considered a hindrance, actually contains rich information about an object's precise intra-frame trajectory.
- Simultaneously addressing blind deblurring and image matting, termed "deblatting," can effectively untangle the image formation process to reveal continuous object motion.
- Incorporating physically justifiable prior knowledge, such as piecewise quadratic curves for object trajectories, is crucial for robustly extracting accurate motion paths from estimated blur kernels.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.