Add crops = results.crop() dictionary

[ELHoussineT]

Adding function that returns list of dicts containing cropped detections and their labels

        >>> result.get_cropped()
            [
                { 
                    'label': <label> <accuracy> 
                    'img': <cropped np image> 
                }, 
                { 
                    'label': <label> <accuracy> 
                    'img': <cropped np image> 
                }, 
                ...
            ]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhancements to image cropping and results output in the YOLOv5 model's inference display method.

📊 Key Changes

• Added a crops list to gather crop data for images during inference.
• Updated the image cropping logic to support conditional saving and added the cropped image data to the crops list.
• Changed the crop method to return the crops list and support optional saving with a specified directory.

🎯 Purpose & Impact

• 🔍 Clarity in Crop Outputs: Users now get detailed information about each crop, improving transparency regarding inference results.
• 💾 Flexible Saving Options: The ability to choose whether to save cropped images or not adds flexibility for different user needs.
• 🔄 Data Accessibility: By returning the crops list, downstream applications can easily utilize the cropping results for further processing or analysis.

[github-actions]

👋 Hello @ELHoussineT, thank you for submitting a 🚀 PR! To allow your work to be integrated as seamlessly as possible, we advise you to:

• ✅ Verify your PR is up-to-date with origin/master. If your PR is behind origin/master an automatic GitHub actions rebase may be attempted by including the /rebase command in a comment body, or by running the following code, replacing 'feature' with the name of your local branch:

git remote add upstream https://github.com/ultralytics/yolov5.git
git fetch upstream
git checkout feature  # <----- replace 'feature' with local branch name
git rebase upstream/master
git push -u origin -f

• ✅ Verify all Continuous Integration (CI) checks are passing.
• ✅ Reduce changes to the absolute minimum required for your bug fix or feature addition. "It is not daily increase but daily decrease, hack away the unessential. The closer to the source, the less wastage there is." -Bruce Lee

[glenn-jocher]

@ELHoussineT thanks for the PR! We can't accept this as-is because it is duplicating code and functionality, you should instead update the existing functionality to return the values you want. Details are below:

Existing display method handles all Detections object tasks like cropping:

yolov5/models/common.py

Lines 367 to 385 in fad57c2

	def display(self, pprint=False, show=False, save=False, crop=False, render=False, save_dir=Path('')):
	for i, (im, pred) in enumerate(zip(self.imgs, self.pred)):
	str = f'image {i + 1}/{len(self.pred)}: {im.shape[0]}x{im.shape[1]} '
	if pred.shape[0]:
	for c in pred[:, -1].unique():
	n = (pred[:, -1] == c).sum() # detections per class
	str += f"{n} {self.names[int(c)]}{'s' * (n > 1)}, " # add to string
	if show or save or render or crop:
	annotator = Annotator(im, pil=not self.ascii)
	for *box, conf, cls in reversed(pred): # xyxy, confidence, class
	label = f'{self.names[int(cls)]} {conf:.2f}'
	if crop:
	save_one_box(box, im, file=save_dir / 'crops' / self.names[int(cls)] / self.files[i])
	else: # all others
	annotator.box_label(box, label, color=colors(cls))
	im = annotator.im
	else:
	str += '(no detections)'

Existing .crop() method that you should simply update with your proposed functionality here:

yolov5/models/common.py

Lines 411 to 415 in fad57c2

	def crop(self, save_dir='runs/detect/exp'):
	save_dir = increment_path(save_dir, exist_ok=save_dir != 'runs/detect/exp', mkdir=True) # increment save_dir
	self.display(crop=True, save_dir=save_dir) # crop results
	LOGGER.info(f'Saved results to {save_dir}\n')

Thank you!

[ELHoussineT]

Hi @glenn-jocher

Sure thing. I added the logic into existing functions (029401d).

I structured it in away that does not introduce radical change. The default save value is False in crop function; this would ensure that the code will not break for people used to the current version of the crop function.

[glenn-jocher]

@ELHoussineT great, thanks for the updates, I will review these tomorrow!

The python crop return feature has been requested before so I think this is a good idea.

[ELHoussineT]

@glenn-jocher Sure thing. I was actually quite surprised you didn't have this before.

[glenn-jocher]

/rebase

[glenn-jocher]

@ELHoussineT PR is merged. Thank you for your contributions to YOLOv5 🚀 and Vision AI ⭐

[glenn-jocher]

PyTorch Hub tutorial updated with Usage example in https://docs.ultralytics.com/yolov5/tutorials/pytorch_hub_model_loading

Cropped Results

Results can be returned and saved as detection crops:

results = model(imgs)  # inference

crops = results.crop(save=True)  # cropped detections dictionary

[ELHoussineT]

Thanks Glenn for sanitizing this.