A lightweight high-performance algorithm named Lighter-Faster-YOLO based on YOLOv8n was proposed in this study. Firstly, the algorithm replaced the conventional convolutions in the original C2f module with the improved depthwise separable partial convolution(DSPConv), which reduced redundant computations and memory access, thereby extracting network features more efficiently. Secondly, it substituted the C2f module before the spatial pyramid pooling-fast(SPPF) with the efficient multi-scale attention(EMA) module, which improved performance with low computational overhead. Finally, a high-level feature fusion pyramid network(HS-FPN) was adopted as the new neck network to enhance the effect of feature fusion and reduce the amount of computation. The results showed that the average precision of the proposed algorithm for grape leaf disease detection reached 93.0%. Compared with the YOLOv8n algorithm, the number of parameters and floating-point operations of the proposed algorithm were reduced by 66.34% and 35.80%, respectively. In comparison with current mainstream lightweight object detection algorithms such as Faster R-CNN and YOLOv5n, the improved Lighter-Faster-YOLO algorithm exhibited superior performance. It could effectively reduce the number of parameters and model complexity, thereby lowering computational costs and facilitating deployment on intelligent detection instruments.