Paper Overview
Research area: Computer Vision Authors: Md. Abdullah Mandal, Saad Ahmed, Md. Khalid Syfullah Published: 2026-09-09 arXiv: 2609.10469Summary
Automated plant disease diagnosis is increasingly deployed on farmer-held devices, but public benchmarks are dominated by a small set of non-native crops, region-specific datasets are rarely validated by domain experts, and architectures reaching competitive accuracy carry parameter budgets unsuited to low-cost hardware. This paper proposes 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 healthy and diseased states of radish, potato, and pointed gourd, collected in the Bogura and Nilphamari districts of Bangladesh.Key Contributions
- AgroVisNet architecture: grouped bottleneck residual blocks with sequential channel and spatial attention, multi-scale depthwise blocks, and a dual-pooling classification head, totaling only 290,572 trainable parameters.
- BD-PlantDX benchmark: 12,432 expert-validated field images across 12 crop disease classes from Bangladesh.
- Performance: 99.52% test accuracy and weighted F1 on BD-PlantDX, exceeding all six ImageNet-pretrained lightweight backbones evaluated under an identical protocol, while using 8.7–16.8x fewer parameters and 1.3–8.5x fewer multiply-accumulate operations.
- Deployment readiness: quantizes to a 0.46 MB full-integer network with only a 0.22 percentage-point accuracy loss, classifying an image in 8.40 ms on a single CPU.
- Robustness and interpretability: accuracy across five random seeds is 99.57 ± 0.10%; a ten-variant ablation isolates each component's contribution; the same architecture transfers without redesign to two independently collected datasets at 98.71% and 99.05% accuracy; Grad-CAM shows predictions rely on lesion-bearing leaf regions rather than background cues.
Original 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.Link: https://arxiv.org/abs/2609.10469
--- *Auto-collected on 2026-09-11*