Regarding the Results of Single Image Testing

[chycxyzd]

@jinlinyi Firstly, thank you for sharing this wonderful work. I tried to test on a single image using the demo you provided, and I found that the test results did not match my expectations. The specific details are as follows:
I use monitoring equipment to take pictures of farmland for a period of four days, with one image taken at a fixed time each day. Therefore, the camera positions, parameters, etc. of the four images remain consistent. However, using these four images for testing yielded inconsistent results, as shown in the following figure:

img1：

img2：

img3:

img4:

In addition, I only changed the focal length of the camera (reducing it by 23 times) and took a new image for testing. The results are as follows:

From the results, it can be seen that the predicted focal length is basically the same as the image before adjusting the focal length.

I am very confused, where is the problem? Can you provide some ideas for solving the problem?