Input image resolution for yolov5 medium model (640x640)

[jelena2712]

Hello everyone,

I am considering the benefits of capturing images from the camera in full HD and then resizing them to the resolution required for object detection. I am interested in understanding how much detail is preserved through this process. Is it perhaps more beneficial to directly capture the image at the resolution that will be used for inference? I am particularly concerned about preserving a high level of detail.

In my case, it is crucial to retain as many details as possible.

Thank you, I would really appreciate any feedback.

Best regards,
Jelena