QCell: Recombining and Aligning Cell Queries for Overlapping Instance Segmentation
QCell targets the hard part of microscopy cell segmentation: separating instances where cells overlap and boundaries fade.
The paper proposes a query-based model that recombines latent instance representations and aligns overlapping cell queries with a contrastive objective. It also introduces an Organoid dataset benchmark for overlapping cell segmentation. The authors report state-of-the-art results across multiple benchmarks, including +2.2 AP and +2.7 AJI on ISBI2014. Accepted at BMVC 2026, with code, models, and dataset linked from the paper. HF Daily Papers' note
The paper proposes a query-based model that recombines latent instance representations and aligns overlapping cell queries with a contrastive objective. It also introduces an Organoid dataset benchmark for overlapping cell segmentation. The authors report state-of-the-art results across multiple benchmarks, including +2.2 AP and +2.7 AJI on ISBI2014. Accepted at BMVC 2026, with code, models, and dataset linked from the paper. HF Daily Papers' note
score 4