Original title: A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition
The system looks at photos of tomato plants and identifies diseases and pests. The researchers compare three ways of finding objects in images, paired with different image-reading networks. They also use local and global labels, which mark details in a small area and across the whole image, plus extra training examples to improve recognition and reduce false alarms. The abstract says the system recognizes nine types of diseases and pests in challenging surroundings.
Imagine checking a tomato plant while getting dinner ready. A leaf may look different because of where it is, how far along the problem is, or what is around it, so one quick glance can be misleading. In this paper, checking the plant is the image detector’s job, while comparing different ways to inspect the photos helps the researchers find a suitable approach.
Farmers need to spot plant problems early enough to treat them and limit economic losses. This work tests image detectors on tomato plants photographed in place, including scenes where diseases and pests vary in location and infection status.
The abstract does not report exact accuracy results, comparisons between the detector combinations, or how well the system performs outside its tomato dataset. It also does not describe field deployment or show that using the detector leads to earlier treatment or lower economic losses.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition — Alvaro Fuentes, Sook Yoon, Sang Ryong Kim et al. — Sensors — doi:10.3390/s17092022
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.