Published May 15, 2020
| Version darknet_yolo_v4_pre
Software
Open
AlexeyAB/darknet: YOLOv4 pre-release
Creators
- Alexey
- Joseph Redmon
- Stefano Sinigardi
- cyy1
- Tino Hager
- Vinjn Zhang2
- Muhammad Maaz3
- IlyaOvodov
- Philip Kahn4
- Josh Veitch-Michaelis5
- Aymeric Dujardin6
- duohappy
- acxz
- John Aughey
- Enes Özipek
- Jud White7
- David Smith8
- Aven
- Tiago Koji Castro Shibata9
- Mosè Giordano10
- Galileo Daras
- HagegeR
- Bartek Gąsiorzewski
- Ali Babaei
- Heethesh Vhavle11
- Evert Arends
- DongChan Cho
- Chih-Hsu Lin12
- Antek Baranski13
- 7FM
- 1. Nowhere
- 2. NVIDIA
- 3. @Confiz
- 4. http://www.velociraptorsystems.com
- 5. Liverpool John Moores University
- 6. @stereolabs
- 7. Dell
- 8. Vanderbilt University
- 9. Microsoft
- 10. @UCL-RITS
- 11. Aeva Inc | Carnegie Mellon University
- 12. Baylor College of Medicine
- 13. Vid Inc
Description
YOLOv4 consists of: https://arxiv.org/abs/2004.10934
- Backbone: CSPDarknet53
- Neck: SPP, PAN
- Head: YOLOv3
Download:
cfg-file: https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v3_optimal/yolov4.cfg
weights-file: https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v3_optimal/yolov4.weights
pre-trained weights-file for training: https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v3_optimal/yolov4.conv.137
Accuracy (AP) / Speed (FPS):
YOLOv4 (416x416 batch=1 FP16) - 32 FPS on Jetson Xavier AGX by using tkDNN+TensorRT: https://github.com/ceccocats/tkDNN
Size Darknet FPS (avg) tkDNN TensorRT FP32 FPS tkDNN TensorRT FP16 FPS tkDNN TensorRT FP16 batch=4 FPS Speedup 320 100.6 116 202 423 4.2x 416 82.5 103 162 284 3.5x 512 69.7 91 134 206 2.9x 680 53.6 62 100 150 2.8xFiles
AlexeyAB/darknet-darknet_yolo_v4_pre.zip
Files
(8.2 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:0ca215babafa55a32fa17a7280810004
|
8.2 MB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/AlexeyAB/darknet/tree/darknet_yolo_v4_pre (URL)