Crop and Weed Detection Data with Bounding Boxes
Dataset of annotated agricultural images, intended for training object detection models to distinguish crops and weeds.
Description
This dataset contains 1300 pictures sesame plantations and various weeds. Each image is 512x512 pixels in size and has been annotated manually with Bounding Boxes according to the YOLO format. The images were obtained from real photographs and enhanced using techniques such as ImageDataGenerator to enrich the initial volume.
What is this dataset for?
- Training object detection models for smart agriculture
- Testing Computer Vision Algorithms to Separate Crops from Weeds
- Create automated crop monitoring applications in the field
Can it be enriched or improved?
Yes. The dataset can be enriched with more plant varieties and light conditions. It is also possible to convert YOLO annotations into other formats such as Pascal VOC or COCO. Manual re-annotation could improve quality for specific use cases (type of crop, region, season, etc.).
🔎 In summary
🧠 Recommended for
- Computer vision data scientists
- Agricultural detection projects
- Embedded AI demonstrations
🔧 Compatible tools
- YoloV5
- Roboflow
- OpenCV
- LabelImg
💡 Tip
To improve performance in real conditions, add images captured at different times and from a variety of angles.
Frequently Asked Questions
Can this dataset be used with Yolov8?
Yes, the annotations are already in YOLO format and can be used directly with YOLOv5, v6, v7, and v8.
Can we convert this dataset to COCO format?
Yes, it is possible to convert YOLO annotations into COCO format using scripts or tools like Roboflow or FiftyOne.
Does this dataset cover different types of crops?
No, it focuses primarily on sesame, but can be extended to other crops by adding new data.
