* autoshape update for PIL greyscale inputs

* autoshape update for PIL greyscale inputs

* Change optimizer parameters group method

* Add torch nn

* Change isinstance method(torch.Tensor to nn.Parameter)

* parameter freeze fix, PEP8 reformat

* freeze bug fix

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* W&B feature improvements

This PR add:
* Class to id labels. Now, the caption of bounding boxes will display the class name and the class confidence score.
* The project name is set to "Yolov5" and the run name will be set to opt.logdir

* cleanup

* remove parenthesis on caption

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* Add wandb metric logging and bounding box debugging

* Improve formatting, readability

* Remove mutliple path for init, improve formatting

* Add wandb params

* Remove typecasting in bbox coordinates and reformat

* Cleanup

* add wandb to requirements.txt

* minor updates to test.py

* general reorg

* reduce --log-imgs to 10

* clean wandb import

* reverse wandb import assert

* add except AssertionError to try import

* move wandb init to all global ranks

* replace print() with logger.info()

* replace print() with logger.info()

* move wandb.init() bug fix

* project PosixPath to basename bug fix

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* torch 1.7.0 compatibility updates

* add inference verification

* Move generated *.json annotations into runs/test

* list and string robust

* Adding --output and --save-conf options to test.py

* Update help fields

* Update test.py

* Make arguments and comments uniform with test.py

* Remove previous and print save_dir on finish

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* README Update

* Update README.md

* README Update

* Update README.md

* Precision-Recall Curve feature update

* sentinel value update

* add contrast to conf/class against bbox rectangle color

* added feature to save conf in txt labels

* Update general.py

* Update detect.py

* Update detect.py

* Update detect.py

save_conf variable removed as only used once, so we can use opt.save_conf instead.

* Update detect.py

* Update detect.py

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

Co-authored-by: swsong <swsong@stratosphere.mobi>

* fix/hyper

* Hyp giou check to train.py

* restore general.py

* train.py overwrite fix

* restore general.py and pep8 update

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* comment

* fix parsing

* fix evolve

* folder

* tqdm

* Update train.py

* Update train.py

* reinstate anchors into meta dict

anchor evolution is working correctly now

* reinstate logger

prefer the single line readout for concise logging, which helps simplify notebook and tutorials etc.

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

img size checks are warnings rather than errors, so current implementation allows improperly formed model inputs.

* initial commit

* Update general.py

Indent update

* Update general.py

refactor duplicate code

* 200 dpi

* added option to not open on save

* Update detect.py

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

* Fix increment_dir to use run_xxx for logdir 

Rerunning train.py with logdir in the form of `run_xxx` causes index slicing in
```
n = max([int(x[len(dir):x.find('_') if '_' in Path(x).name else None]) for x in d]) + 1  # increment
```
to underflow.

* Replace find with rfind