Computer vision is a field that enables computers to perceive and understand the world through images and videos. It has been extensively studied in various aspects of precision agriculture, including autonomous harvesting robots, plant phenotyping, crop yield estimation, plant pest and disease detection, animal welfare assessment, and so on. Autonomous harvesting robots have been developed to fulfill crop picking tasks, replacing manual picking, using a computer vision system functioning as the eye of the robot. Computer vision is applied to detect fruits/vegetables and locate their three-dimensional positions. Due to the uncertainties in the field, such as varying illumination and severe occlusions, and the lack of publicly available image datasets, computer vision for autonomous harvesting robots remains a challenging field.

Phenotype refers to the measurable characteristics of a plant, such as leaf shape and area. Phenotype describes the relationship between genotype and environment on a plant’s measurable characteristics. Measuring plant phenotype accurately is important. Example areas of research include approaches to segment plant parts accurately and methods to reconstruct plant parts.

Yield estimation is important for farmers to improve field management. With the help of precise yield maps, farmers can utilize prescription maps for variable-rate irrigation and fertilizer application. Also, yield estimation allows farmers to plan ahead for storage and sales. Current yield estimation is mainly done by manual counting, which is labor-intensive and inaccurate. Computer vision may improve the efficiency and accuracy of yield estimation.

Computer vision can also be applied to plant pest and disease detection, and animal welfare assessment, e.g. detection of different animal behavior, poses, and body conditions
This Research Topic issue will cover varied applications of computer vision in precision agriculture, including autonomous harvesting robots, plant phenotyping, crop yield estimation, plant pest and disease detection, and animal welfare assessment. Topics of interest include (but are not limited to):
• Publicly available image dataset for agriculture. These datasets could be used for training and testing fruit/vegetable detection algorithms or plant-part segmentation algorithms.
• Fruit/vegetable detection. Preferably, the development of a general, real-time method that could detect all kinds of crops in the field.
• Fruit/vegetable localization. Preferably, the development of a general method that could precisely locate all kinds of crops in the fields.
• Plant-part segmentation or reconstruction.
• Precise crop yield estimation.
• Plant pest and disease detection.
• Animal welfare assessment using computer vision
• Health monitoring of farmland and water conservancy facilities

Submit your idea

Impact

Frontiers in Robotics and AI

Participating Journals

Impact