Weak event detection is a key step to improve the accuracy of first-arrival picking in microseismic monitoring. Traditional STA/LTA and AIC methods have limited detection capability under low signal?to?noise ratio conditions. This paper proposes a microseismic event detection method based on the YOLOv8 object detection model, which transforms event detection into a candidate?window object detection task. Experimental results show that YOLOv8 achieves an overall mAP50 of 0.948 on the test set, with a precision of 0.841, recall of 0.917, and mAP50 of 0.956 for weak events, and a false?detection rate of 0% on both types of negative samples. On this basis, a YOLOv8?detection?box?guided AIC joint first?arrival picking strategy is proposed. The detection box locates candidate event windows, within which the AIC function is computed for accurate picking, and multi?trace continuity constraints are further applied to improve the stability of the first?arrival picks. This method provides an effective technical solution for automated processing of low?SNR microseismic data.