Search
← Back to Discoveries | 2026-08-20 synthesized

AI model spots rapeseed field weeds fast on cheap hardware

Crop Improvement

Farmers walking oilseed rape fields could soon point a low-cost camera at a patch and get instant, precise weed identification instead of guessing which spots need herbicide.

Scientists trained a computer vision model to tell weeds apart from rapeseed plants in photos, then trimmed it down so it runs quickly on small, inexpensive devices instead of needing a powerful computer. The slimmed-down model actually got more accurate than the original while working nearly 30% faster, correctly identifying weeds and crop about 94% of the time at a speed of over 25 frames per second.

Key Findings

1

The improved MMB-YOLO model reached 93.8% precision, 93.8% recall, 95.8% mAP50, and 76.0% mAP50-95, all higher than the baseline YOLOv13n model.

2

Computational cost (GFLOPs) dropped by 29.7% thanks to a MobileNetV3 backbone and a new lightweight MBConv detection head.

3

On an edge device the model detected weeds at an average of 25.3 frames per second, fast enough for real-time field use.

chevron_right Technical Summary

Researchers built a lightweight AI model that spots weeds growing among rapeseed plants with over 93% accuracy while running fast enough on cheap portable hardware to guide real-time weeding equipment.

description

Abstract Preview

Original paper

Lightweight detection of companion weeds in rapeseed based on improved YOLOv13n

Abstract To solve the problems in accurately and efficiently detecting weeds in rapeseed fields under complex conditions, this paper proposes an improved MMB-YOLO model based on YOLOv13n. This mode...

open_in_new Read full abstract

Abstract copyright held by the original publisher.

hub This connects to 10 other discoveries — Rapeseed crop-improvement, invasive-species, urban-ecology +1 more 5 related articles

Species Mentioned

Was this useful?

mail Weekly plant science — one email, Saturdays.

Share: X/Twitter Reddit
arrow_forward Next Discovery

Gene editing removes 97% of celiac-triggering proteins from bread wheat

It could mean that people with celiac disease — roughly 1 in 100 worldwide — may one day safely eat bread made from real wheat, without sacrificing the taste...