基于锚索状态图像识别的退锚机器人作业可行性判别
|
|
刘治翔,庄源,谢春雪
|
Operation feasibility discrimination of anchor withdrawing robot based on anchor cable state image recognition
|
|
Zhixiang LIU,Yuan ZHUANG,Chunxue XIE
|
|
| 表1 YOLOv8目标检测模型参数设置 |
| Table 1 Parameter settings of YOLOv8 object detection model |
|
| 参数 | 数值 | 参数 | 数值 |
|---|
| lr0 | 0.01 | hsv_h | 0.015 | | lrf | 0.01 | hsv_s | 0.7 | | 最终学习率 | 0.000 1 | hsv_v | 0.4 | | 余弦学习率 | False | deterministic | True | | 优化器 | ADAM | translate | 0.1 | | 标签平滑参数 | 0 | scale | 0.5 | | 正则化 | 0 | epochs | 300 | | Weight_decay | 5×10-4 | imgsz | 640 | | warmup_epochs | 3 | flipud | 0 | | warmup_momentum | 0.8 | fliplr | 0.5 | | warmup_bias_lr | 0.1 | half | True | | box | 0.05 | conf | 0.5 | | cls | 0.5 | line_thickness | 8 | | cls_pw | 1 | iou | 0.7 | | obj | 1 | max_det | 300 | | obj_pw | 1 | anchor_t | 4 | | iou_t | 0.2 | split | val |
|
|
|