Dataset · dronefreak
YOLO-formatted ExDark low-light object detection
dronefreak/ExDark
Mirror of the well-known ExDark low-light photography dataset, reformatted for YOLO.
What's inside
Unofficial ExDark dataset repackaging from the University of Malaya, containing between 10,000 and 100,000 images with English annotations. It gathers shots captured in darkness and low-light conditions with annotations for object detection. The entire folder and file structure is adapted to a standard format compatible with YOLO.
What you can do with it
- Training YOLO detectors for night surveillance and autonomous driving
- Evaluating the robustness of computer vision models in the dark
- Fine-tuning object detectors under difficult lighting conditions
- Tasks
- Object detection
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 27.9K
- Likes
- 9
Models trained on it
- dronefreak/exdark-rfdetr-nanoObject detection 46
- dronefreak/exdark-rfdetr-smallObject detection 45
- dronefreak/exdark-yolov8nObject detection 40
- dronefreak/exdark-yolo11xObject detection 39
- dronefreak/exdark-yolo11nObject detection 37
- dronefreak/exdark-rfdetr-mediumObject detection 34
- dronefreak/exdark-yolov8mObject detection 33
- dronefreak/exdark-yolo26sObject detection 32
How to load it
from datasets import load_dataset
ds = load_dataset("dronefreak/ExDark")