What is health detection

One of the most critical parts in the vineyard is early detection of diseases in the plants. That is why the BACCHUS project is trying to do early detection of diseases utilizing the sensors on our inspection robot in combination with AI.

Why do we detect diseases

BACCHUS partners have developed a segmentation algorithm and trained it on photos from grapevine plants taken on the field, depicting both healthy and infected plants. The U-NET algorithm was selected for higher accuracy on the disease localization and was evaluated offline.

Segmentation model architecture (left) and its validation on the localization of the grapevine diseases (right)

How about weed detection

While commercial devices able to detect weeds on bare soil already exist, weed detection at more advanced crop growth stage remains a challenging problem. Until now, most studies deal with image spatial information to discriminate weeds from crop, using a colour or multispectral imaging sensor.

In BACCHUS we provide real time weed infestation maps, as the input parameters will be updated dynamically during the robot movement in the field, independently of the row structure and growth stage, as well as information on the weed species involved, leading to significant progress in site-specific weed management.

The following images illustrate weed detection at different cultivation periods either using the side camera and the front camera of the inspection platform.

Weed detection validation from unknown images using the robot’s side camera
Intra-row weed detection. Color illustration of height distribution per frame
Project Coordinator
Prof. Zoe Doulgeri
Automation & Robotics Lab
Aristotle University of Thessaloniki
Department of Electrical & Computer Engineering
Thessaloniki 54124, Greece

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 871704.

The website reflects only the view of the author(s) and the Commission is not responsible for any use that may be made of the information it contains.

Bacchus Project BACCHUS Project
© 2020 - 2022 All rights reserved