Deployment-aware HDBO-R optimisation of a lightweight ECA-MBConv-Net for tea leaf disease classification

Manual visual assessment of tea leaf diseases is labour-intensive, subjective, and difficult to scale across large plantation environments. Accurate and deployable image-based recognition is important for timely tea crop management. To address the accuracy–efficiency trade-off in mobile tea leaf disease classification, this study proposes a lightweight deployment-aware framework based on ECA-MBConv-Net and a hybrid Dung Beetle Optimizer–RIME algorithm, termed HDBO-R. The proposed network integrates Efficient Channel Attention into MBConv-style depthwise separable blocks to enhance disease-relevant channel representation with limited computational overhead. HDBO-R searches architectural and training configurations using a deployment-aware objective that jointly considers macro-F1, parameter count, FLOPs, and CPU inference latency. Experiments were conducted on three datasets: the authors’ Sri Lankan tea leaf disease dataset, an external Indian tea leaf disease dataset, and a Taiwan tomato leaf dataset for cross-crop evaluation. On the SLTea dataset, ECA-MBConv-Net achieved 99.33% macro-F1 with 86.2K parameters and 11.75 ms CPU latency. The final HDBO-R optimised model achieved 98.90% macro-F1 while reducing parameters by 22.47% to 66.9K, FLOPs by 4.91% to 235.3M, and CPU latency by 7.06% to 10.92 ms. Across the three datasets, HDBO-R reduced CPU latency by 4.19–7.06% while preserving competitive classification performance. The final FP32 TensorFlow Lite model was deployed on three Android smartphones, with mean model-only latency ranging from 37.40 ms to 92.74 ms. These results demonstrate that deployment-aware optimisation can produce compact, efficient, and mobile-ready models for practical tea leaf disease recognition.