Summary
Diagnosing plant diseases through smartphones can help farmers, especially in places without expert advice or good internet access. The authors created a small and efficient AI model called AgroVisNet that can identify diseases in radish, potato, and pointed gourd plants using photos taken in fields in Bangladesh. They also put together a large, expert-checked image dataset called BD-PlantDX to train and test the model. AgroVisNet is very accurate, works fast on simple devices, and focuses on the actual damaged parts of leaves to make its decisions. The model also performs well on other similar plant datasets without needing changes.
convolutional neural networkplant disease classificationlightweight modeldataset validationBD-PlantDXattention mechanismdepthwise convolutionsmodel quantizationimage classificationGrad-CAM
Authors
Md. Abdullah Mandal, Saad Ahmed, Md. Khalid Syfullah
Abstract
Automated plant disease diagnosis is increasingly deployed on farmer-held devices in regions where agronomic expertise is scarce and network connectivity is unreliable. Three obstacles limit its practical value: public benchmarks are dominated by a small set of non-native crops, region-specific datasets are rarely validated by domain experts, and the architectures that reach competitive accuracy carry parameter budgets that are unsuited to low-cost hardware. We propose AgroVisNet, a compact convolutional network trained from scratch, together with BD-PlantDX, an expert-validated benchmark of 12,432 field images spanning 12 classes of radish, potato and pointed gourd in healthy and diseased states, collected across the Bogura and Nilphamari districts of Bangladesh. AgroVisNet couples grouped bottleneck residual blocks carrying sequential channel and spatial attention with multi-scale depthwise blocks and a dual-pooling classification head, reaching 290,572 trainable parameters. On BD-PlantDX the model attains 99.52% test accuracy and 99.52% weighted F1, exceeding all six ImageNet-pretrained lightweight backbones evaluated under an identical protocol while using 8.7 to 16.8 times fewer parameters and 1.3 to 8.5 times fewer multiply-accumulate operations. Exported for deployment, the model quantises to a 0.46 MB full-integer network at a 0.22 percentage-point accuracy cost and classifies an image in 8.40 ms on a single CPU. Across five random seeds accuracy remains at 99.57 +- 0.10%, a ten-variant ablation isolates the contribution of each component, and the same architecture transfers without redesign to two independently collected datasets at 98.71% and 99.05% accuracy. Grad-CAM evidence indicates that predictions rest on lesion-bearing leaf regions rather than on background cues.