Published March 2, 2020
| Version darknet_yolo_v3_optimal
Software
Open
AlexeyAB/darknet: Yolo v3 optimal
Authors/Creators
- Alexey
- Joseph Redmon
- Stefano Sinigardi
- cyy1
- Tino Hager
- Vinjn Zhang2
- IlyaOvodov
- Philip Kahn3
- Josh Veitch-Michaelis4
- Aymeric Dujardin5
- duohappy
- acxz
- John Aughey
- Jud White6
- David Smith7
- Aven
- Tiago Koji Castro Shibata8
- Mosè Giordano9
- Galileo Daras10
- HagegeR
- Bartek Gąsiorzewski
- Puneet Kohli11
- Timon12
- Nadeen Udantha
- Marcin Kmiec
- Heethesh Vhavle13
- DongChan Cho
- Chih-Hsu Lin14
- Antek Baranski15
- 7FM
- 1. Nowhere
- 2. NVIDIA
- 3. http://www.velociraptorsystems.com
- 4. Liverpool John Moores University
- 5. @stereolabs
- 6. Dell
- 7. Vanderbilt University
- 8. Microsoft
- 9. @UCL-RITS
- 10. @helium
- 11. Leia Inc
- 12. MESO Digital Interiors
- 13. Aeva Inc | Carnegie Mellon University
- 14. Baylor College of Medicine
- 15. Vid Inc
Description
Features:
- fusion blocks: FPN, PAN, ASFF, BiFPN
- network modules: ResNet, CPS, SPP, RFB
- network architecture search: CSPResNext50, CSPDarknet53, SpineNet49, EfficientNetB0, MixNet-M
- activations: SWISH, MISH
- other features: weighted-[shortcut], Sigmoid scaling (Scale-sensitivity), Label smoothing, Optimal hyper parameters, Dynamic mini batch size for random shapes, Squeeze-and-excitation, Grouped convolution, MixConv (grouped [route]), Elastic-module
- data augmentation: MixUp, CutMix, Mosaic
- losses: MSE, GIoU, CIoU, DIoU
- detection layers: [yolo] (fixed iou_thresh), [Gaussian_yolo]
- detection on video (sequence of frames) - layers: [crnn] (convolutional-RNN), [conv_lstm] (Convolutional LSTM)
Files
AlexeyAB/darknet-darknet_yolo_v3_optimal.zip
Files
(8.1 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:703fae7e0d475030fa15816e6305952d
|
8.1 MB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/AlexeyAB/darknet/tree/darknet_yolo_v3_optimal (URL)