diff --git a/PaddleCV/PaddleVideo/metrics/detections/README.md b/PaddleCV/PaddleVideo/metrics/detections/README.md new file mode 100644 index 0000000000000000000000000000000000000000..24fa08a4162b0e3b9bc30755788f1c5d26cc98aa --- /dev/null +++ b/PaddleCV/PaddleVideo/metrics/detections/README.md @@ -0,0 +1,13 @@ +## ActivityNet 指标计算 + +- ActivityNet数据集的具体使用说明可以参考其[官方网站](http://activity-net.org) + +- 下载指标评估代码,请从[ActivityNet Gitub repository](https://github.com/activitynet/ActivityNet.git)下载 + +- 计算精度指标 + + ```cd ActivityNet/Evaluation``` + + ```python get_detection_performance.py ./data/activity_net.v1-3.min.json $Test_Result``` + + 其中Test_Result是运行测试程序test.py输出的json格式的文件。 diff --git a/PaddleCV/PaddleVideo/models/ctcn/README.md b/PaddleCV/PaddleVideo/models/ctcn/README.md index 136cf989396d162498ca395a52bd770afc7f2480..4d247d764155a6fb04e3f61eeddf652ba6128d79 100644 --- a/PaddleCV/PaddleVideo/models/ctcn/README.md +++ b/PaddleCV/PaddleVideo/models/ctcn/README.md @@ -59,6 +59,8 @@ C-TCN的训练数据采用ActivityNet1.3提供的数据集,数据下载及准 - 若未指定`--weights`参数,脚本会下载已发布模型[model](https://paddlemodels.bj.bcebos.com/video_detection/ctcn.tar.gz)进行评估 +- 运行上述程序会将测试结果保存在json文件中,使用ActivityNet官方提供的测试脚本,即可计算MAP。具体计算过程请参考[指标计算](../../metrics/detections/README.md) + 当取如下参数时,在ActivityNet1.3数据集下评估精度如下: | score\_thresh | nms\_thresh | soft\_sigma | soft\_thresh | MAP |