Icevision

Added
- pytorch 1.7 support, all dependencies updated
- tutorial with hard negative samples
- ability to skip record while parsing

Changed
- show_preds visual improvement

Added
- Cache records after parsing with the new parameter `cache_filepath` added to `Parser.parse` (504)
- Added `pretrained: bool = True` argument to both faster_rcnn and mask_rcnn `model()` methods. (516)
- new class `EncodedRLEs`
- all masks get converted to `EncodedRLEs` at parsing time

Changed
- Removed warning on autofixing masks
- RLE default counts is now COCO style
- renamed `Mask.to_erle` to `Mask.to_erles`

Changed
- updated matplotlib and ipykernel minimum version for colab compatibility

IMPORTANT
- **Switched from poetry to setuptools**

Added
- Function `wandb_img_preds` to help logging bboxes to wandb
- wandb as a soft dependency
- Template code for `parsers.SizeMixin` if `parsers.FilepathMixin` is used
- Get image size without opening image with `get_image_size`
- Ability to skip record while parsing with `AbortParseRecord`
- Autofix for record: `autofix_records` function and `autofix:bool` parameter added to `Parser.parse`
- Record class and mixins, `create_mixed_record` function to help creating Records
- InvalidDataError for BBox
- Catches InvalidDataError while parsing data

Changed
- **Breaking:** Unified `parsers.SizeMixin` functions `image_width` and `image_height` into a single function `image_width_height`
- Rename Parser `SizeMixin` fields from `width` `height` to `image_width` `image_height`

Deleted
- Removed `CombinedParser`, all parsing can be done with the standard `Parser`

Added
- Efficientdet now support empty annotations

Changed
- Returns float instead of dict on `FastaiMetricAdapter`

Changed

- Updates fastai2 to the final release version