Knowledge Distillation Facilitates the Lightweight and Efficient Plant Diseases Detection Model

计算机科学 植物病害 分割 蒸馏 人工智能 机器学习 目标检测 航程(航空) 数据挖掘 工程类 生物 航空航天工程 生物技术 有机化学 化学
作者
Qianding Huang,Xingxin Wu,Qi Wang,Xinyu Dong,Liang Liu,Xin Wu,Yangyang Gao,Ge‐Fei Hao
出处
期刊:Plant phenomics [American Association for the Advancement of Science]
卷期号:5 被引量:1
标识
DOI:10.34133/plantphenomics.0062
摘要

Plant disease diagnosis in time can inhibit the spread of the disease and prevent a large-scale drop in production, which benefits food production. Object detection-based plant disease diagnosis methods have attracted widespread attention due to their accuracy in classifying and locating diseases. However, existing methods are still limited to single crop disease diagnosis. More importantly, the existing model has a large number of parameters, which is not conducive to deploying it to agricultural mobile devices. Nonetheless, reducing the number of model parameters tends to cause a decrease in model accuracy. To solve these problems, we propose a plant disease detection method based on knowledge distillation to achieve a lightweight and efficient diagnosis of multiple diseases across multiple crops. In detail, we design 2 strategies to build 4 different lightweight models as student models: the YOLOR-Light-v1, YOLOR-Light-v2, Mobile-YOLOR-v1, and Mobile-YOLOR-v2 models, and adopt the YOLOR model as the teacher model. We develop a multistage knowledge distillation method to improve lightweight model performance, achieving 60.4% mAP@ .5 in the PlantDoc dataset with small model parameters, outperforming existing methods. Overall, the multistage knowledge distillation technique can make the model lighter while maintaining high accuracy. Not only that, the technique can be extended to other tasks, such as image classification and image segmentation, to obtain automated plant disease diagnostic models with a wider range of lightweight applicability in smart agriculture. Our code is available at https://github.com/QDH/MSKD.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
风清扬应助不知干嘛采纳,获得10
1秒前
红炉点血完成签到,获得积分10
2秒前
4秒前
輕瘋完成签到,获得积分10
6秒前
7秒前
Tonald Yang发布了新的文献求助10
7秒前
强公子完成签到,获得积分10
8秒前
8秒前
rh完成签到,获得积分10
8秒前
11秒前
Ivy完成签到,获得积分10
13秒前
13秒前
落后十八完成签到,获得积分10
15秒前
vbnn发布了新的文献求助30
15秒前
LHT发布了新的文献求助10
15秒前
落后乐荷完成签到,获得积分10
16秒前
piaoaxi完成签到 ,获得积分10
16秒前
妮妮完成签到,获得积分10
17秒前
前进中完成签到,获得积分10
18秒前
韭菜盒子发布了新的文献求助10
19秒前
19秒前
Q22发布了新的文献求助10
19秒前
鲁丁丁完成签到 ,获得积分10
20秒前
jason完成签到,获得积分10
21秒前
mike2012完成签到 ,获得积分10
21秒前
KINGAZX完成签到 ,获得积分10
24秒前
LHT完成签到,获得积分10
25秒前
研友_VZG7GZ应助韭菜盒子采纳,获得10
25秒前
CLY完成签到,获得积分10
27秒前
Q22完成签到,获得积分10
27秒前
weiyayayayayaya完成签到,获得积分10
27秒前
CAOHOU应助虚心的芹采纳,获得10
28秒前
折花几慕应助虚心的芹采纳,获得20
28秒前
折花几慕应助虚心的芹采纳,获得20
28秒前
大胆砖头完成签到 ,获得积分10
28秒前
celia完成签到 ,获得积分10
28秒前
Beyond095完成签到,获得积分10
29秒前
搜集达人应助Q22采纳,获得10
31秒前
31秒前
万能图书馆应助夏傥采纳,获得10
33秒前
高分求助中
Semantics for Latin: An Introduction 1155
Genomic signature of non-random mating in human complex traits 1000
Plutonium Handbook 1000
Three plays : drama 1000
Robot-supported joining of reinforcement textiles with one-sided sewing heads 600
SPSS for Windows Step by Step: A Simple Study Guide and Reference, 17.0 Update (10th Edition) 500
Multimodal injustices: Speech acts, gender bias, and speaker’s status 400
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 遗传学 基因 物理化学 催化作用 冶金 细胞生物学 免疫学
热门帖子
关注 科研通微信公众号,转发送积分 4106192
求助须知:如何正确求助?哪些是违规求助? 3644100
关于积分的说明 11542941
捐赠科研通 3351096
什么是DOI,文献DOI怎么找? 1841209
邀请新用户注册赠送积分活动 907950
科研通“疑难数据库(出版商)”最低求助积分说明 825090