ExDark is a dataset for low-light object detection. It contains annotated images for training YOLO models. The dataset was enhanced with additional images and labels.
Use Cases
- Train YOLO models to detect objects in low-light images using the provided bounding box labels.
- Benchmark object detection algorithm performance in challenging lighting conditions.
- Augment existing object detection datasets with low-light scenarios.
- Fine-tune pre-trained models on the enhanced image set for specialized low-light applications.
Strengths
- Enhanced version includes additional images and labels beyond the original ExDark collection.
- Dataset is formatted specifically for YOLO model training with a provided dataset.yaml file.
Limitations
- Exact number of images, rows, and classes is unknown.
- Original dataset may have a limited number of object categories compared to general-purpose datasets.
Provenance
- Source
- Kaggle
- Collection Method
- Enhanced version of the original ExDark (Exclusively Dark) dataset.