Ultralytics Hub: Merge Dataset feature

[adolfusadams]

The Ultralytics Hub platform has been extremely beneficial to my workflow. However, there is a feature that it lacks, that is, the ability to merge datasets.

When a user clicks on "Train Model" there should or could be some checkboxes that would allow for multi-selecting the datasets. See attached image as an example. Proceeding to train model would merge the selected datasets.

[pderrenger]

@adolfusadams thank you for your valuable feedback and for sharing your experience with Ultralytics HUB! 😊

The idea of merging datasets directly within the platform and selecting multiple datasets for training is indeed a compelling feature that could enhance the flexibility and efficiency of the workflow. We appreciate your suggestion and the detailed example you provided.

Currently, the platform does not support merging datasets directly. However, you can manually merge datasets by combining their respective directories and updating the dataset YAML file accordingly. Here’s a quick example of how you can do this:

1. Combine Image Directories: Merge the image directories from both datasets into a single directory structure.
2. Update YAML File: Create a new YAML file that includes paths to the combined image directories.

For instance, if you have dataset1 and dataset2, your combined YAML might look like this:

# Combined dataset YAML
path: ../datasets/combined_dataset
train: images/train
val: images/val
test: images/test

# Classes
names:
  0: class1
  1: class2
  # Add all classes from both datasets

After preparing your combined dataset, you can upload it to Ultralytics HUB and proceed with training.

We will certainly pass your feature request to our development team for consideration in future updates. Your input is invaluable in helping us improve the platform to better meet user needs.

If you encounter any issues or have further suggestions, please feel free to share. Thank you for being an active member of our community! 🚀

For more detailed guidance on dataset preparation, you can visit our Ultralytics HUB Datasets Documentation.