Minimum image resolution

Dear team, all,
what’s the minimum image resolution to be more accurate as possible.
It seems that using D1 720 x 576 Deepstack miss “people” when they are far from the camera

I think you are asking, what are the smallest size of person that can be detected by the model (relative to the size of the image). This question will not have a definite answer as detection is dependent on many factors (eg illumination of scene, any obscurations), and also as new more accurate models come into deepstack, detection will improve. Your best approach is to experiment, then adjust the location of you cameras to be nearer the subject