Merge DatasetManager / GenericDataset / OCRDatasetManager / OCRDataset classes
Merged
requested to merge merge-datasetmanager-genericdataset-ocrdatasetmanager-ocrdataset-classes into main
All threads resolved!
Closes #120 (closed)
Merge request reports
Activity
changed milestone to %DAN-P4: Improve data loading and preprocessing
added P2 label
assigned to @starride
So far I merged:
-
GenricDataset
andOCRDataset
→OCRDataset
. This class is a simpletorch.utils.data.Dataset
that loads images & labels for a specific set -
DatasetManager
andOCRDatasetManager
→OCRDatasetManager
. This class handles generic parameters (charset, tokens, batch_size, data transforms), and creates the three datasets + dataloaders + datasamplers. It is then used indan/manager/training.py
.
-
- Resolved by Solene Tarride
added 6 commits
-
37f984fe...fdaf48f4 - 4 commits from branch
main
- a3266590 - Simplify dataset classes
- b6a3e8d1 - rename function
-
37f984fe...fdaf48f4 - 4 commits from branch
Please register or sign in to reply