版本一
The first model is only used to run the training process, prompt word: dztx, There is smoke surrounding the slowly moving object in the camera.
The data set only has 30 pairs