Published May 12, 2018
| Version 0.3
Software
Open
fizyr/keras-retinanet 0.3
Creators
- Hans Gaiser1
- Maarten de Vries1
- Ashley Williamson2
- Valeriu Lacatusu
- András
- Yann Henon3
- jjiun
- Cristian Gratie
- Mihai Morariu
- Enrico Liscio4
- Rodrigo Meira de Andrade
- Pedro Conceição
- hannesedvartsen
- Wudi Fang
- Mike Clark
- Iver Jordal5
- Max Van Sande
- Ori226
- Pierre Dérian6
- Muhammed Kocabas
- srslynow
- Guillaume Erhard
- Eduardo Ramos7
- Diego
- 1. @fizyr
- 2. @INPStarfall @LCAS
- 3. @netrasys
- 4. @delftrobotics
- 5. AIA Science
- 6. CEA Tech
- 7. @paradigmadigital
Description
Changes since last release
- Implement DenseNet, VGG backbones.
- Add option to freeze backbone layers.
- Add logging of evaluation to tensorboard.
- Add pretty colors for 80 classes.
- Fix batch_size > 1 issues.
- Refactor model outputs (should hopefully stay like this now).
- Simplified training by splitting into "training model" and "inference model".
- Add structure for backbone specific functions (such as
load_model
). - Encode regression as x1/y1/x2/y2 offsets (increases mAP to 0.350, previously 0.345).
- Use
nearest
upsampling method.
Credits to: @vidosits @cgratie @DiegoAgher @eduramiba @GuillaumeErhard @Muhannes @hgaiser @iver56 @jjiunlin @srslynow @de-vri-es @Ori226 @pedroconceicao @pderian @rodrigo2019 @lvaleriu @yhenon
Files
fizyr/keras-retinanet-0.3.zip
Files
(2.2 MB)
Name | Size | Download all |
---|---|---|
md5:47d3d0265f844ff3bd6c785aef1512f0
|
2.2 MB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/fizyr/keras-retinanet/tree/0.3 (URL)