300 high-resolution RGB images containing 4,432 manually annotated grape clusters across 5 distinct wine grape varieties. The dataset provides instance-level annotations including bounding boxes and pixel-wise binary masks for Cabernet Franc, Cabernet Sauvignon, Chardonnay, Merlot, and Syrah cultivars in field conditions.
Use Cases
- Train instance segmentation models like Mask R-CNN using the provided binary mask PNG files.
- Develop automated yield estimation algorithms by counting instances via the YOLO-formatted bounding box coordinates.
- Benchmark object detection performance across different grape varieties using the cultivar-specific labels.
Strengths
- 300 images featuring 4,432 individual grape cluster instances.
- Covers 5 specific grape cultivars: Cabernet Franc, Cabernet Sauvignon, Chardonnay, Merlot, and Syrah.
- Includes dual annotation formats: YOLO-compatible bounding boxes and binary segmentation masks in PNG format.
- Captured in natural vineyard environments with varying lighting, shadows, and leaf occlusions.