c687d5c129
reorganize train initialization steps
2020-09-04 12:25:53 -07:00
bc1fd13a7a
gsutil cp hyp evolution bug fix ( #876 )
2020-09-03 20:29:08 -07:00
44cdcc7e0b
hyp['anchors'] evolution update
2020-09-03 12:54:22 -07:00
ffe9eb4238
remove if platform.system() == 'Darwin' checks
2020-09-03 11:41:33 -07:00
23a7f29e2d
Update hyp.finetune.yaml on 306 VOC generations result
2020-09-03 01:20:30 -07:00
0a0675da12
Merge remote-tracking branch 'origin/master'
2020-09-03 01:07:48 -07:00
354a97e4ef
Create sotabench.py
2020-09-03 01:06:59 -07:00
d8274d0434
Fix results_file not renaming ( #903 )
2020-09-03 00:47:50 -07:00
281d78c105
Update train.py ( #902 )
...
* Update train.py with simplified ckpt names
* Return default hyps to hyp.scratch.yaml
Leave line commented for future use once mystery of best finetuning hyps to apply becomes clearer.
* Force test_batch*_pred.jpg replot on final epoch
This will allow you to see predictions final testing run after training completes in runs/exp0
2020-09-02 15:08:43 -07:00
d68179953d
Update export.py
2020-09-02 13:23:29 -07:00
6f3db5e662
Remove autoanchor and class checks on resumed training ( #889 )
...
* Class frequency not calculated on resuming training
Calculation of class frequency is not needed when resuming training.
Anchors can still be recalculated whether resuming or not.
* Check rank for autoanchor
* Update train.py
no autoanchor checks on resume
Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2020-09-02 09:05:31 -07:00
c5969f79f5
add support for DWConv fuse operation ( #898 )
...
* the value should be a list to unpacking
* filter the None labels
* update readme to remove apex dependency
* add support for DWConv fuse operation
2020-09-01 22:18:42 -07:00
d3f9bf2bb7
Update datasets.py
2020-09-01 17:02:47 -07:00
901243c780
Fix relative path for VOC download ( #880 )
...
Cleaned up code
2020-08-31 16:06:10 -07:00
3e04d20c7d
Merge remote-tracking branch 'origin/master'
2020-08-31 11:07:00 -07:00
f06e2d518c
opt.image_weights bug fix ( #885 )
2020-08-31 11:05:37 -07:00
1e15aad6f9
Add InfiniteDataLoader class ( #876 )
...
* Add InfiniteDataLoader
Only initializes at first epoch. Saves time.
* Moved class to a better location
2020-08-31 11:01:25 -07:00
69ff781ca5
opt.img_weights bug fix ( #885 )
2020-08-31 10:33:07 -07:00
987c226849
seaborn plotting correlogram addition
2020-08-29 15:13:58 -07:00
08e97a2f88
Update hyperparameters to add lrf, anchors
2020-08-28 14:58:43 -07:00
9776e70988
torch.ops.torchvision.nms ( #860 )
...
Don't load the entire torchvision library just for nms when the function is already in the torch library.
2020-08-27 11:13:41 -07:00
0c01afc390
Update Dockerfile to upgrade pip
2020-08-26 14:44:43 -07:00
62e037419f
Update Dockerfile FROM nvcr.io/nvidia/pytorch:20.08-py3
2020-08-26 00:44:34 -07:00
31d6c389f2
Update general.py plot_evolution()
2020-08-25 23:08:14 -07:00
1e7674d4dd
Update export.py --img-size comment
2020-08-25 23:07:22 -07:00
a21bd0687c
Update train.py forward simplification
2020-08-25 13:48:03 -07:00
455f7b8f76
Update general.py ( #823 )
...
Fixes #822
`init_seeds` from `torch_utils` import is being overwritten by function `init_seeds` in `general.py`
2020-08-25 13:01:17 -07:00
5a7d79fbe6
Update general.py bbox_iou() with eps improvements ( #736 )
...
eps reduced to 1e-9 for additional precaution.
2020-08-25 12:56:39 -07:00
82ed33a052
generalize webcam detection with opt.source.isnumeric() (yolov3 #1462 )
2020-08-25 11:07:38 -07:00
09402a2174
torch.from_tensor() bug fix
2020-08-25 03:14:17 -07:00
4fb8cb353f
Update export.py with v3.0 Hardswish() support ( #831 )
2020-08-24 21:59:26 -07:00
4d7f222f73
Update export.py with v3.0 Hardswish() support
2020-08-24 21:47:49 -07:00
8666bc507e
Create evolve.sh for infinite hyp evolution
2020-08-24 20:27:28 -07:00
a8751e50de
model.fuse() fix for export.py ( #827 )
2020-08-24 19:27:54 -07:00
5f07782f56
Update requirements.txt for coremltools kmeans quantization
2020-08-22 20:50:53 -07:00
71209a6099
exportable Hardswish() implementation
2020-08-22 16:59:31 -07:00
fd71fe8451
Update hyp.finetune.yaml
2020-08-22 16:18:53 -07:00
f7e075fe93
gsutil cp yaml when evolving to GCP bucket
2020-08-22 15:47:39 -07:00
9dbaa1bbbe
Update plot_results()
2020-08-22 15:32:05 -07:00
83dc540b1d
remove ema.ema hasattr(ema, 'module') check
2020-08-22 15:18:39 -07:00
1309400e6b
Update hyp.finetune.yaml on VOC results
2020-08-22 15:05:47 -07:00
fb05681a04
Update requirements.txt to coremltools==4.0b3
2020-08-22 14:22:15 -07:00
d0f98c0095
Update yolo.py with model_info(verbose=False)
2020-08-21 16:20:45 -07:00
b9851222bf
Merge remote-tracking branch 'origin/master'
2020-08-21 15:03:09 -07:00
c171e45cda
Update test.py with set() ( #806 )
2020-08-21 15:03:03 -07:00
ea5e97fb9b
FP16 NMS ( #814 )
2020-08-21 14:44:25 -07:00
b42e8a531b
optimize imports
2020-08-20 21:17:40 -07:00
4447f4b937
--resume to same runs/exp directory ( #765 )
...
* initial commit
* add weight backup dir on resume
2020-08-20 18:24:33 -07:00
5e0b90de8f
CIoU nan bug fix ( #736 )
2020-08-20 10:37:55 -07:00
fb4fc8cd02
Fix ema attribute error in DDP mode ( #775 )
...
* Fix ema error in DDP mode
* Update train.py
Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2020-08-18 09:52:21 -07:00
Glenn Jocher
NanoCode012
Naman Gupta
tkianai
rafale77
Willie Maddox