Qidian213/deep_sort_yolov3

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1675 stars · 586 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A real-time multi-person tracker that combines YOLOv3 object detection with the Deep SORT tracking algorithm, implemented in Python with TensorFlow/Keras. It detects and tracks pedestrians (and optionally other COCO classes) in video streams or from a camera at roughly 11 fps on a GTX 1060.

Use cases

• track multiple people in a video in real time
• pedestrian tracking with yolo and deep sort
• count and follow people across camera frames
• multi-object tracking demo with tensorflow
• detect and track persons from webcam feed
• build a people tracking pipeline for surveillance video

When to choose

• you need a simple, ready-to-run YOLOv3 + Deep SORT baseline in TensorFlow 1.x
• you want to track pedestrians in real time on a modest GPU
• you want a reference implementation to swap in a different detector

When to avoid

• you need a maintained or modern implementation - the author recommends PyTorch alternatives
• you want to train models or use recent TensorFlow/PyTorch versions
• you need high accuracy state-of-the-art MOT performance (FairMOT, CenterTrack, etc.)
• you need to track non-person classes without modifying code

Facets

application · maturity abandoned

computer-vision image-processing machine-learning deep-learning computer-vision artificial-intelligence python object-tracking multi-object-tracking yolo deep-sort pedestrian-tracking tensorflow keras real-time video gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Qidian213/deep_sort_yolov3")