作业:基于 RTMDet 的气球检测
背景:熟悉目标检测和 MMDetection 常用自定义流程。
任务:
- 基于提供的 notebook,将 cat 数据集换成气球数据集;
- 按照视频中 notebook 步骤,可视化数据集和标签;
- 使用MMDetection算法库,训练 RTMDet 气球目标检测算法,可以适当调参,提交测试集评估指标;
- 用网上下载的任意包括气球的图片进行预测,将预测结果发到群里;
- 按照视频中 notebook 步骤,对 demo 图片进行特征图可视化和 Box AM 可视化,将结果发到群里
- 需提交的测试集评估指标(不能低于baseline指标的50%)
- 目标检测 RTMDet-tiny 模型性能 不低于65 mAP。
1.可视化数据集和标签
数据集
便签
2.测试集评估:
Evaluate annotation type *bbox*
DONE (t=0.11s).
Accumulating evaluation results...
DONE (t=0.01s).
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.734
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.843
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.829
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.373
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.872
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.240
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.770
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.822
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.683
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.914
06/11 07:53:36 - mmengine - INFO - bbox_mAP_copypaste: 0.734 0.843 0.829 0.000 0.373 0.872
06/11 07:53:36 - mmengine - INFO - Epoch(test) [13/13] coco/bbox_mAP: 0.7340 coco/bbox_mAP_50: 0.8430 coco/bbox_mAP_75: 0.8290 coco/bbox_mAP_s: 0.0000 coco/bbox_mAP_m: 0.3730 coco/bbox_mAP_l: 0.8720 data_time: 0.0152 time: 0.0754
3.网络气球图像预测:
4.可视化分析
可视化 backbone 输出的 3 个通道
可视化 neck 输出的 3 个通道
查看 neck 输出的最小输出特征图的 Grad CAM
查看 neck 输出的最大输出特征图的 Grad CAM
5.Log&Checkpoint
见 log与 checkpoint 文件夹