PuzLM: Solving Jigsaw Puzzles with Sequence-to-Sequence Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Elkin, Gur, Shahar, Ofir Itzhak, Ben-Shahar, Ohad
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912962691727360
author Elkin, Gur
Shahar, Ofir Itzhak
Ben-Shahar, Ohad
author_facet Elkin, Gur
Shahar, Ofir Itzhak
Ben-Shahar, Ohad
contents Square jigsaw puzzles are typically solved by visually matching piece images to recover the original layout. This work introduces PuzLM, an alternative perspective that recasts jigsaw reassembly as a discrete sequence-to-sequence (Seq2Seq) problem, inspired by natural language representations. We design an efficient puzzle quantization procedure that transforms each piece into a short sequence of discrete tokens, enabling the direct application of standard Seq2Seq language models as powerful jigsaw solvers. Our approach demonstrates that accurate puzzle reconstruction can be achieved through purely symbolic reasoning over discrete representations, improving state-of-the-art performance even on puzzles with eroded boundaries or missing pieces.
format Preprint
id arxiv_https___arxiv_org_abs_2511_06315
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PuzLM: Solving Jigsaw Puzzles with Sequence-to-Sequence Language Models
Elkin, Gur
Shahar, Ofir Itzhak
Ben-Shahar, Ohad
Computer Vision and Pattern Recognition
Square jigsaw puzzles are typically solved by visually matching piece images to recover the original layout. This work introduces PuzLM, an alternative perspective that recasts jigsaw reassembly as a discrete sequence-to-sequence (Seq2Seq) problem, inspired by natural language representations. We design an efficient puzzle quantization procedure that transforms each piece into a short sequence of discrete tokens, enabling the direct application of standard Seq2Seq language models as powerful jigsaw solvers. Our approach demonstrates that accurate puzzle reconstruction can be achieved through purely symbolic reasoning over discrete representations, improving state-of-the-art performance even on puzzles with eroded boundaries or missing pieces.
title PuzLM: Solving Jigsaw Puzzles with Sequence-to-Sequence Language Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.06315