Cityperson json label to yolo label

[swaingotnochill]

❔Question

I am trying to evaluate the model( pretrained weights) on Cityperson dataset. I wish to convert the labels into yolo format, but can't do it. Do you have any idea how should I approach this? Also, is there any other thing I should take care for the evaluation?

Additional context

I already have the labels for the cityperson dataset. The question is I wish to replace all the pedestrian, rider, etc with a single class and also am not sure of the object class Id.
Any suggestions is valuable.

[github-actions]

👋 Hello @swaingotnochill, thank you for your interest in 🚀 YOLOv5! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://www.ultralytics.com or email Glenn Jocher at [email protected].

Requirements

Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), testing (test.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

@swaingotnochill you should export your labels into YOLO format and create a citypersons.yaml file for this dataset. Then you can train and test normally using train.py and test.py. See Train Custom Data tutorial to get started:

YOLOv5 Tutorials

• Train Custom Data  🚀 RECOMMENDED
• Tips for Best Training Results  ☘️ RECOMMENDED
• Weights & Biases Logging  🌟 NEW
• Supervisely Ecosystem  🌟 NEW
• Multi-GPU Training
• PyTorch Hub  ⭐ NEW
• ONNX and TorchScript Export
• Test-Time Augmentation (TTA)
• Model Ensembling
• Model Pruning/Sparsity
• Hyperparameter Evolution
• Transfer Learning with Frozen Layers  ⭐ NEW
• TensorRT Deployment

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[NguyenThanhBinh98]

Question

I am trying to evaluate the model( pretrained weights) on Cityperson dataset. I wish to convert the labels into yolo format, but can't do it. Do you have any idea how should I approach this? Also, is there any other thing I should take care for the evaluation?

Additional context

I already have the labels for the cityperson dataset. The question is I wish to replace all the pedestrian, rider, etc with a single class and also am not sure of the object class Id. Any suggestions is valuable.

[NguyenThanhBinh98]

Hi swaingotnochill . Can you share with me the labeled cityperson dataset . I want train Yolov5 with that

[glenn-jocher]

@NguyenThanhBinh98 @swaingotnochill yes it would be great to create a Cityperson.yaml for everyone to be able to autodownload this dataset.

Please see our ✅ Contributing Guide to get started.

[swaingotnochill]

@NguyenThanhBinh98 @glenn-jocher I don't have the dataset as of now. But the dataset is available in https://www.cityscapes-dataset.com/ . I have wrote the script for converting the cityperson labels to yolo lables. If that's ok, maybe I can create an PR for that.

[glenn-jocher]

@swaingotnochill yes great! If you can create a Cityscapes.yaml similar to our other dataset yamls and start populating it what you have, that will help get this started.

[swaingotnochill]

@glenn-jocher I don't think that is possible. I had a conversation with a CityScape person, and they don't allow free download of the datasets. Their dataset is available only for research/education purpose. So , they don't have an Open Download Link. When I downloaded the dataset, I had to submit my University Student details to be verified. This is a troublesome process, but that is how they operate :(

[glenn-jocher]

@swaingotnochill yes this should be ok. We have the same setup in place with the xView dataset. The users download xView zips manually to /datasets, and then run python train.py --data xView.yaml, and we automatically extract and convert to YOLO format and start training.

Also we are unaffected by dataset licensing since we are not an end user. It would be up to end users (like yourself) to download CityScape zips first and ensure they are following license guidelines.

yolov5/data/xView.yaml

Lines 1 to 9 in 99a45ba

	# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
	# xView 2018 dataset https://challenge.xviewdataset.org
	# -------- DOWNLOAD DATA MANUALLY from URL above and unzip to 'datasets/xView' before running train command! --------
	# Example usage: python train.py --data xView.yaml
	# parent
	# ├── yolov5
	# └── datasets
	# └── xView ← downloads here

[codename-cn]

@swaingotnochill you don't necessarily have to provide the dataset itself. However, you could have provided the script for the conversion.

[glenn-jocher]

@codename-cn Sharing your script for converting Cityperson labels to YOLO format would indeed be very helpful for the community! If you could create a pull request with your script, that would be fantastic. This way, others can benefit from your work once they have access to the dataset themselves. 😊👍