After creating a new project, you will come to the next phase to upload images (training dataset) for the project
Load Image
To upload the training dataset
- Select the folder icon (1)
- Browse to the dataset directory (2)
- Select the current folder (3) to start loading the dataset into ODHUB
Data Validation
Once the dataset is uploaded, you can preview the image in the middle panel.
The next thing you need to do is to make sure the dataset is sufficient for model training:
- Step 1: Review the Image and Label File (A) dialogue to confirm the dataset has sufficient image and label files. If the dataset is not sufficient, you can use the Labelling Tool to add more labels to the dataset until it meets all requirements.
- Step 2: Once the dataset is all checked, click Validate Data. This will generate a label map file and compare it with the labelled files to verify all label categories (C)
- Step 3: Ensure the Map File (B) is all checked.
- Step 4: Once data is GOOD, press Send data to engine (E) to go to the training phase