1747 Commits

Author SHA1 Message Date
Glenn Jocher
ca19df5f7f
Add autobatch feature for best batch-size estimation (#5092)
* Autobatch

* fix mem

* fix mem2

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update train.py

* print result

* Cleanup print result

* swap fix in call

* to 64

* use total

* fix

* fix

* fix

* fix

* fix

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Cleanup printing

* Update final printout

* Update autobatch.py

* Update autobatch.py

* Update autobatch.py
2021-10-25 13:56:13 +02:00
Ayush Chaurasia
692be757b6
W&B: Media panel fix (#5317) 2021-10-25 12:06:12 +02:00
Zhiqiang Wang
3d897986c7
Small fixes to docstrings (#5313)
* Minor fixes of the git checkout new branch

* Use em dash to quote

* Revert the change of git checkout

* Maybe we should up-to-date with the upstream/master?
2021-10-24 22:05:34 +02:00
Cristi Fati
fee83c1634
Weights download script minor improvements (#5213)
* Add nano model to the download list

* Add possibility to also download the "*6" models

* Cleanup

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-23 14:20:26 +02:00
Jebastin Nadar
e0c3f42de7
Uncomment OpenCV 4.5.4 requirement in detect.py (#5305) 2021-10-23 13:40:34 +02:00
Glenn Jocher
79d8f1f678 Created using Colaboratory 2021-10-22 21:19:23 +02:00
Glenn Jocher
b760acec11 Created using Colaboratory 2021-10-22 21:11:29 +02:00
Glenn Jocher
441b47c443
More informative EarlyStopping() message (#5303) 2021-10-22 20:02:19 +02:00
jdfr
15e8c4c15b
take EXIF orientation tags into account when fixing corrupt images (#5270)
* take EXIF orientation tags into account when fixing corrupt images

* fit 120 char

* sort imports

* Update local exif_transpose comment

We have a local inplace version that is faster than the official as the image is not copied. AutoShape() uses this for Hub models, but here it is not important as the datasets.py usage is infrequent (AutoShape() it is applied every image).

* Update datasets.py

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-20 21:35:47 +02:00
jdfr
db3bbdde0e
autosplit: take image files with uppercase extensions into account (#5269)
* take image files with uppercase extensions into account in autosplit

* case fix

* Refactor implementation

Removes additional variable (capital variable names are also only for global variables), and uses the same methodology as implemented earlier in datasets.py L409.

* Remove redundant rglob characters

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-20 19:25:33 +02:00
Glenn Jocher
a18b0c36cd
Update rebase.yml with workflows permissions (#5255)
Update token with workflows permissions
2021-10-19 12:02:15 +02:00
dependabot[bot]
fd3e2ad8ef
Bump actions/stale from 3 to 4 (#5253)
Bumps [actions/stale](https://github.com/actions/stale) from 3 to 4.
- [Release notes](https://github.com/actions/stale/releases)
- [Changelog](https://github.com/actions/stale/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/stale/compare/v3...v4)

---
updated-dependencies:
- dependency-name: actions/stale
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-10-19 11:37:54 +02:00
dependabot[bot]
aa26504137
Bump actions/cache from 1 to 2.1.6 (#5252)
Bumps [actions/cache](https://github.com/actions/cache) from 1 to 2.1.6.
- [Release notes](https://github.com/actions/cache/releases)
- [Commits](https://github.com/actions/cache/compare/v1...v2.1.6)

---
updated-dependencies:
- dependency-name: actions/cache
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-10-19 11:31:24 +02:00
dependabot[bot]
fb12faafa3
Bump cirrus-actions/rebase from 1.4 to 1.5 (#5251)
Bumps [cirrus-actions/rebase](https://github.com/cirrus-actions/rebase) from 1.4 to 1.5.
- [Release notes](https://github.com/cirrus-actions/rebase/releases)
- [Commits](https://github.com/cirrus-actions/rebase/compare/1.4...1.5)

---
updated-dependencies:
- dependency-name: cirrus-actions/rebase
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-10-19 11:28:52 +02:00
Zhiqiang Wang
8e87396705
Add dependabot for GH actions (#5250) 2021-10-19 11:23:30 +02:00
Glenn Jocher
4f5e274eee
Update rebase.yml (#5245) 2021-10-19 09:46:05 +02:00
Glenn Jocher
752d97df2c
Add on_fit_epoch_end callback (#5232)
* Add `on_fit_epoch_end` callback

* Add results to train

* Update __init__.py
2021-10-18 14:28:22 +02:00
Glenn Jocher
13f7275555
Update/inplace ops (#5233)
* Clip Objects365 autodownload labels (#5214)

Fixes out of bounds labels that seem to affect ~10% of images in dataset.

* Inplace ops
2021-10-18 14:24:48 +02:00
Glenn Jocher
0000334a6f
Update Objects365.yaml val count (#5212) 2021-10-15 12:40:40 -07:00
Glenn Jocher
991c654e81
Autofix duplicate label handling (#5210)
* Autofix duplicate labels

PR changes duplicate label handling from report error and ignore image-label pair to report warning and autofix image-label pair. 

This should fix this common issue for users and allow everyone to get started and get a model trained faster and easier than before.

* sign fix

* Cleanup

* Increment cache version

* all to any fix
2021-10-15 12:32:08 -07:00
Farley Lai
fc3606420d
Update Objects365.yaml to include the official validation set (#5194)
* Update Objects365.yaml

Download the official Objects365 validation set and convert the labels

* Enforce 4-space indent, reformat and cleanup

* shorten list comprehension

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-14 20:48:47 -07:00
Glenn Jocher
1c9f710b46
Add hyp.scratch-med.yaml (#5196)
* Add hyp.scratch-med.yaml

* Update hyp.scratch-med.yaml
2021-10-14 20:40:30 -07:00
Glenn Jocher
6b31cfd793
Highlight contributors in README (#5173)
* Highlight contributors in README

* Adjust Contact section

* Update README.md
2021-10-14 12:49:16 -07:00
Glenn Jocher
7ee5aed0b3
Improved check_suffix() robustness to '' and "" (#5192)
* Improved check_suffix() robustness to `''` and `""`

* Cleanup
2021-10-14 12:00:39 -07:00
NauchtanRobotics
0be58f16f0
Adjust legend labels for classes without instances (#5174)
* legend-labels Adjust legend labels for classes without instances.

* #5158 Re-indexed series names: only classes containing data.

* #5158 Re-indexed series names: only classes containing data.

* Cleanup

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-14 11:57:00 -07:00
Ayush Chaurasia
1172ac15fb
W&B: DDP fix (#5176)
* fix dpp with wandb disabled

* PyCharm reformat

* Fix

* fix

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-13 20:11:04 -07:00
Glenn Jocher
e520e820f1
Update autodownload fallbacks to v6.0 assets (#5177) 2021-10-13 19:58:38 -07:00
Ayush Chaurasia
d0bfeb3980
W&B: fix DDP with wandb disabled (#5163)
* fix dpp with wandb disabled

* PyCharm reformat

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-13 16:17:15 -07:00
Glenn Jocher
a346926996
Add class filtering to LoadImagesAndLabels() dataloader (#5172)
* Add train class filter feature to datasets.py

Allows for training on a subset of total classes if `include_class` list is defined on datasets.py L448:
```python
        include_class = []  # filter labels to include only these classes (optional)
```

* segments fix
2021-10-13 15:48:54 -07:00
Glenn Jocher
b754525e99
Check 'onnxruntime-gpu' if torch.has_cuda (#5087)
* Check `'onnxruntime-gpu' if torch.has_cuda`

* fix indent
2021-10-12 22:25:38 -07:00
Glenn Jocher
20a809dee5
Update stale.yml (#5156) 2021-10-12 17:54:56 -07:00
Glenn Jocher
26784afd99
Use YOLOv5n for CI testing (#5154) 2021-10-12 08:42:26 -07:00
Yoni Chechik
34da872ab6
fix tf conversion in new v6 models (#5153)
* fix `tf` conversion in new v6 (#5147)

* sort imports

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-12 08:38:54 -07:00
Glenn Jocher
956be8e642
YOLOv5 release v6.0 (#5141)
* Update P5 models

* Update P6 models

* Update with GFLOPs and Params

* Update with GFLOPs and Params

* Update README

* Update

* Update README

* Update

* Update

* Add times

* Update README

* Update results

* Update results

* Update results

* Update hyps

* Update plots

* Update plots

* Update README.md

* Add nano models to hubconf.py
v6.0
2021-10-11 23:47:18 -07:00
Andrei Ionut Damian
938efe5dcf
update detect.py in order to support torch script (#5109)
* update detect.py in order to support torch script

This change assumes the torchscrip file was previously saved with `export.py`

* update `detect.py` for torchscript support

Simple update for torchscript support. Assumes the torchscript file has been generated with `export.py`

* Cleanup

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-11 21:56:06 -07:00
Glenn Jocher
0bf24cf641
Add OpenCV DNN option for ONNX inference (#5136)
* Add OpenCV DNN option for ONNX inference

Usage:

```bash
python detect.py --weights yolov5s.onnx  # ONNX Runtime inference
python detect.py --weights yolov5s.onnx -dnn  # OpenCV DNN inference
```

* DNN prediction to tensor

* Update detect.py
2021-10-11 12:39:20 -07:00
Jebastin Nadar
9d75e42f98
Refactor Detect() anchors for ONNX <> OpenCV DNN compatibility (#4833)
* refactor anchors and anchor_grid in Detect Layer

* fix CI failures by adding compatibility

* fix tf failure

* fix different devices errors

* Cleanup

* fix anchors overwriting issue

* better refactoring

* Remove self.anchor_grid shape check (redundant with self.grid check)

Also PEP8 / 120 line width

* Convert _make_grid() from static to dynamic method

* Remove anchor_grid.to(device)

clone() should already clone to same device as self.anchors

* fix different devices error

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-11 09:58:42 -07:00
Malte Lorbach
153873e9e4
Fix ROOT as relative path (#5129)
* use os.path.relpath instead of relative_to

* use os.path.relpath instead of relative_to

* Remove os.path from val.py

* Remove os.path from train.py

* Update detect.py import to os

* Update export.py import to os

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-11 09:47:24 -07:00
Glenn Jocher
4cf7d487a6
Update plots.py (#5127) 2021-10-11 01:59:22 -07:00
Glenn Jocher
7d37b3c52e
Update val.py pad = 0.0 if task == speed else 0.5 (#5121)
* Update val.py `pad = 0.0 if task == speed else 0.5`

* Cleanup
2021-10-10 23:20:42 -07:00
Glenn Jocher
ba4b79de8b
Update val.py --speed and --study usages (#5120) 2021-10-10 21:15:28 -07:00
Glenn Jocher
4f9718abe6
Pass --device for --task study (#5118) 2021-10-10 15:07:26 -07:00
Jebastin Nadar
a0e15046c3
Fix different devices bug when moving model from GPU to CPU (#5110)
* fix different devices bug

* extend _apply() instead of to() for a general fix

* Only apply if Detect() is last layer

Co-authored-by: Jebastin Nadar <njebastin10@gmail.com>

* Indent fix

* Add comment to yolo.py

* Add comment to common.py

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-10 14:24:05 -07:00
Nan
4a6dfffdaa
Pass LOCAL_RANK to torch_distributed_zero_first() (#5114)
Co-authored-by: qiningonline <qiningonline@gmail.com>
2021-10-09 18:41:50 -07:00
hamid mohammadi
97b6b14abe
Add requests to requirements.txt (#5112)
* Requirements updated

1. Requests added to requirements.txt. That might not be included in all docker base images, adding it to the requirements is safer.
2. Added a minimum version to Pandas. It's a good practice to have versions for all dependencies.

* Sort alphabetically

* Update requirements.txt

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-09 15:51:01 -07:00
Glenn Jocher
136640eee8 Improve doctstring, add usage examples 2021-10-09 15:22:12 -07:00
Glenn Jocher
76259b15b0 Hyp fix 2021-10-09 14:40:42 -07:00
Glenn Jocher
f9c254b207 Merge master 2021-10-09 14:38:28 -07:00
Glenn Jocher
276b674513
Fix SKU-110K HUB: OSError (#5106) 2021-10-09 00:12:59 -07:00
Egor Orel
48b00dbc0d
Re-open IP-camera videostream if disconnected (#5074)
* Re-open IP-camera videostream if disconnected. Provide IP-streaming analysis stability

* Re-open IP-camera videostream if disconnected. Provide IP-streaming analysis stability

* `self.imgs[i] *= 0` bug fix and cleanup

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
2021-10-07 23:26:43 -07:00