SAMannot: A Memory-Efficient, Local, Open-source Framework for Interactive Video Instance Segmentation based on SAM2

TL;DR

SAMannot is an open-source, memory-efficient framework that enables high-fidelity, privacy-preserving video instance segmentation with human-in-the-loop interaction, suitable for research and complex annotation tasks.

Contribution

It introduces a modified SAM2 dependency and a processing layer to reduce resource requirements, enhancing responsiveness and usability for research workflows.

Findings

Verified on animal behavior tracking datasets.

Provides scalable, private, and cost-effective annotation.

Supports research-ready dataset generation in YOLO and PNG formats.

Abstract

Current research workflows for precise video segmentation are often forced into a compromise between labor-intensive manual curation, costly commercial platforms, and/or privacy-compromising cloud-based services. The demand for high-fidelity video instance segmentation in research is often hindered by the bottleneck of manual annotation and the privacy concerns of cloud-based tools. We present SAMannot, an open-source, local framework that integrates the Segment Anything Model 2 (SAM2) into a human-in-the-loop workflow. To address the high resource requirements of foundation models, we modified the SAM2 dependency and implemented a processing layer that minimizes computational overhead and maximizes throughput, ensuring a highly responsive user interface. Key features include persistent instance identity management, an automated ``lock-and-refine'' workflow with barrier frames, and a mask-skeletonization-based auto-prompting mechanism. SAMannot facilitates the generation of research-ready datasets in YOLO and PNG formats alongside structured interaction logs. Verified through animal behavior tracking use-cases and subsets of the LVOS and DAVIS benchmark datasets, the tool provides a scalable, private, and cost-effective alternative to commercial platforms for complex video annotation tasks.