# nwojke/deep_sort

Simple Online Realtime Tracking with a Deep Association Metric

Repository: https://github.com/nwojke/deep_sort
Canonical: https://ross.abutalabs.com/products/deep_sort
Language: Python
License: GPL-3.0
License Family: copyleft
Last push: 2025-03-02T09:55:38+00:00

## Health v2 (maintenance only)
Score: 36/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 9, release rhythm 35, longevity 100
- inputs: {"age_days": 3497, "days_push": 549, "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 6169, forks 1552 (observed 2026-08-28T04:09:36.812811+00:00)

## What it is
A Python implementation of Deep SORT, a multi-object tracking algorithm that extends SORT with a deep appearance descriptor for robust data association. It tracks pedestrians in real time using Kalman filtering and cosine-similarity matching of CNN appearance features.

## Use cases
- track pedestrians across video frames in real time
- run multi-object tracking on MOT16 benchmark sequences
- generate deep appearance features for person re-identification
- associate object detections over time using appearance similarity
- evaluate tracking algorithms on MOT challenge data

## When to choose
- you need real-time multi-object tracking with appearance-based re-identification
- you want a well-known reference implementation of the Deep SORT paper
- you are benchmarking trackers on MOT challenge datasets

## When to avoid
- you need actively maintained code or modern TensorFlow 2 support
- you need tracking for arbitrary object classes rather than pedestrians
- you want a production-ready pipeline with built-in object detection

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, machine-learning, deep-learning
- domain: computer-vision, artificial-intelligence
- platform: python
- tags: multi-object-tracking, deep-sort, person-re-identification, mot-challenge, kalman-filter, tensorflow, algorithms

## Member repositories
- nwojke/deep_sort (main) score 36

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:36.812811+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 2026-08-29T17:47:27.574034+00:00, confidence not recorded.
  - readme: https://github.com/nwojke/deep_sort (fetched 2026-08-28T04:09:36.812811+00:00, sha 8b84482314b3)
- Data as of 2026-08-30T08:39:29.467469+00:00.