The predicted result is represented on what frame

[kankanzheli]

My problem is that this code predicts 15 consecutive images. For example, a video is 30 frames. The detection result of 1-15 frames is cross. Is the detection result of cross displayed on each frame?

I see that in the process of video processing, the first 14 frames of video have no result. So I think the overall judgment of these 15 frames is shown on frame 15. Then frame 16 shows the judgment result of frame (2-16).

So am I right?

If I am right, your code will not have the function of predicting pedestrians. The data extracted by your code (cross) is the image of pedestrians crossing the road and the image of the first five frames before crossing the road. In this way, the result of the training is that the tenth frame (the first five frames + the last ten frames) of the pedestrian will be correctly judged as cross

So can you tell me your answer as soon as possible?