生成的多天气数据集

[wenwang122]

请问在测试的时候对无人机应用的多天气效果，比如说针对于雪天，这所有无人机照片所加的雪天效果都是一模一样的吗，比如说雪花的位置都是固定的吗？
要是我用你的文件iaa_trans.py文件把原始数据集生成雪天效果保存到本地，然后在进行测试，这会导致生成的最后天气效果和你生成的有差距吗？

[Jahawn-Wen]

The added weather effects are not consistent.
Although the author set a fixed seed (for example, seed=35) in the enhancement method of imgaug, this approach does not necessarily guarantee that the weather effect on each picture is exactly the same.

[wincj]

The added weather effects are not consistent. Although the author set a fixed seed (for example, seed=35) in the enhancement method of imgaug, this approach does not necessarily guarantee that the weather effect on each picture is exactly the same.

你好大佬，想请教你我是不是要先运行iaa_trans，他可以直接批量生成所有图片的多天气条件是吗？
然后我才能进行接下来的训练

[zhoucreate]

The added weather effects are not consistent. Although the author set a fixed seed (for example, seed=35) in the enhancement method of imgaug, this approach does not necessarily guarantee that the weather effect on each picture is exactly the same.

你好大佬，想请教你我是不是要先运行iaa_trans，他可以直接批量生成所有图片的多天气条件是吗？ 然后我才能进行接下来的训练

During the model training phase, you don't need to perform iaa_trans.py in advance. train.py already contains the process of generating data in multiple weather conditions.