Strawberry ripeness detection is of great significance for evaluating fruit growth in strawberry orchards. However, the existing detection technology has challenges in mobile terminal deployment due to the large computing amount and large model size. Therefore, an improved lightweight Y010v5 algorithm for strawberry ripeness detection was proposed in this study. To reduces the calculated amount and model scale, Partial Convolution (PConv) is introduced to substitute the normal convolution, which is in the original C3(Cross Stage Partial Bottleneck with 3 convolutions.), and C3_Faster module is introduced to substitute the C3 module in yolov5. SIoU is used as the boundary box loss function to accelerate the regression prediction and improve the positioning accuracy of the object bounding box. Experiments on jetson nano on public dataset show that compared with the previous Y010v5 model, the mAP of the modified model is increased by 1.1%, FLOPs and params are decreased by 17.1% and 17.6%, respectively.