인공지능 환경세팅

YOLO + Object Tracking

노갈이 2022. 11. 30. 06:57
728x90

Author: KangchanRoh

Team: Reinforcement Learning Team @ CAI Lab
Date: 2022/11/30


Benchmark

COCO Benchmark (Real-Time Object Detection)

 

Papers with Code - COCO Benchmark (Real-Time Object Detection)

The current state-of-the-art on COCO is YOLOv7-E6E (36 fps). See a full comparison of 41 papers with code.

paperswithcode.com

MOT17 Benchmark (Multi-Object Tracking)

 

Papers with Code - MOT17 Benchmark (Multi-Object Tracking)

The current state-of-the-art on MOT17 is SMILEtrack. See a full comparison of 28 papers with code.

paperswithcode.com

MOT20 Benchmark (Multi-Object Tracking)

 

Papers with Code - MOT20 Benchmark (Multi-Object Tracking)

The current state-of-the-art on MOT20 is SMILEtrack. See a full comparison of 15 papers with code.

paperswithcode.com

DanceTrack Benchmark (Multi-Object Tracking)

 

Papers with Code - DanceTrack Benchmark (Multi-Object Tracking)

The current state-of-the-art on DanceTrack is MOTRv2. See a full comparison of 11 papers with code.

paperswithcode.com

CroHD Benchmark (Multi-Object Tracking)

 

Papers with Code - CroHD Benchmark (Multiple Object Tracking)

The current state-of-the-art on CroHD is PP-Tracking. See a full comparison of 5 papers with code.

paperswithcode.com

Paper & Code

StrongSORT

StrongSORT_Make_DeepSORT_Great_Again.pdf
6.02MB

OC - SORT

Observation-Centric_SORT_Rethinking_SORT_for_Robust.pdf
7.64MB

ByteTrack

ByteTrack_Multi-Object_Tracking_by_Associating_Every_Detection_Box.pdf
8.24MB

YOLOv5v6 + StrongSORT & OC-SORT & ByteTrack Code

 

GitHub - mikel-brostrom/Yolov5_StrongSORT_OSNet: A collection of real-time multi-camera multi-object SOTA trackers using YOLOv5

A collection of real-time multi-camera multi-object SOTA trackers using YOLOv5 - GitHub - mikel-brostrom/Yolov5_StrongSORT_OSNet: A collection of real-time multi-camera multi-object SOTA trackers u...

github.com

YOLOv5v6 성능

YOLOv7 + StrongSORT Code

 

GitHub - mikel-brostrom/Yolov7_StrongSORT_OSNet: Real-time multi-camera multi-object tracker using YOLOv7 and StrongSORT with OS

Real-time multi-camera multi-object tracker using YOLOv7 and StrongSORT with OSNet - GitHub - mikel-brostrom/Yolov7_StrongSORT_OSNet: Real-time multi-camera multi-object tracker using YOLOv7 and S...

github.com

YOLOv7 성능

FPS Test

  • OS : Ubuntu 20.04
  • CPU : Intel i5-4690 CPU @ 3.50GHz × 4
  • GPU : NVIDIA GeForce GTX 1660
  • RAM : 16GB
Detector & Tracker StrongSORT OC-SORT ByteTrack
YOLOv7 ~7~14 - -
YOLOv7-tiny ~7~21 - -
YOLOv7-X ~7~11 - -
YOLOv7-D6 ~5~10 - -
YOLOv7-E6 ~6~11 - -
YOLOv7-E6E ~5~9 - -
YOLOv7-W6 ~6~14 - -
YOLOv5n ~22~30 ~56 ~71 ~57~71
YOLOv5s ~14~26 ~44~60 ~57~69
YOLOv5m ~7~20 ~35~41 ~39~42
YOLOv5l ~9~16 ~24~26 ~23~26
YOLOv5x ~6~11 ~7~15 ~14~15
YOLOv5n6 ~16~25 ~51~63 ~50~60
YOLOv5s6 ~13~24 ~50~59 ~53~60
YOLOv5m6 ~8~20 ~28~35 ~35~37
YOLOv5l6 ~8~15 ~20~23 ~22~23
YOLOv5x6 ~7~10 ~6~14 ~13~14

 

반응형