A benchmark for jigsaw puzzle solving with irregular, archaeologically-inspired fragment shapes, as presented in CVPR 2026 paper: "The Missing GAP: From Solving Square Jigsaw Puzzles to Handling Real World Archaeological Fragments". GAP contains puzzles derived from natural images that are fragmented using erosion masks, producing non-rectangular pieces with an alpha channel encoding their shape.
The fragment shapes are sampled… See the full description on the dataset page:
https://huggingface.co/datasets/Ofirish/GAP.
