# STVIR/pysot

SenseTime Research platform for single object tracking, implementing algorithms like SiamRPN and SiamMask.

Repository: https://github.com/STVIR/pysot
Canonical: https://ross.abutalabs.com/products/pysot
Language: Python
License: Apache-2.0
License Family: permissive
Topics: tracking, computer-vision, siamese-network
Last push: 2025-06-22T03:59:22+00:00

## Health v2 (maintenance only)
Score: 45/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 28, release rhythm 35, longevity 100
- inputs: {"age_days": 2675, "days_push": 437, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4600, forks 1109 (observed 2026-08-28T04:08:54.834469+00:00)

## What it is
PySOT is a Python research platform by SenseTime for single object visual tracking, implementing algorithms such as SiamRPN, SiamRPN++, DaSiamRPN, SiamFC, and SiamMask on PyTorch. It also includes an evaluation toolkit supporting standard tracking benchmarks like OTB2015, VOT, LaSOT, and UAV123.

## Use cases
- track a single object in video with deep learning
- run SiamRPN or SiamMask tracking models
- evaluate visual trackers on OTB, VOT, LaSOT benchmarks
- research new siamese network tracking algorithms
- demo object tracking from a webcam
- download pretrained tracking model checkpoints

## When to choose
- you need state-of-the-art single object tracking baselines for research
- you want a PyTorch codebase to extend with novel tracking algorithms
- you need standardized evaluation on common tracking benchmarks

## When to avoid
- you need multi-object tracking rather than single object tracking
- you want a production-ready tracking service rather than a research codebase
- you need a framework-agnostic or non-PyTorch solution

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, machine-learning, deep-learning
- domain: computer-vision, machine-learning, deep-learning
- platform: python
- tags: object-tracking, siamese-network, pytorch, siamrpn, siammask, visual-tracking, research, linux, gpu

## Member repositories
- STVIR/pysot (main) score 45

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:54.834469+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:19:47.296235+00:00, confidence not recorded.
  - readme: https://github.com/STVIR/pysot (fetched 2026-08-28T04:08:54.834469+00:00, sha f269036433ee)
- Data as of 2026-08-30T08:39:29.467469+00:00.