Question about Input size of image

[kronee0516]

Now my project want to put 1920 x 1080 sized image size to do the inference
but i see no yolov5 model handling such a image size
is that impossible to do so ? or i can just choose the largest image size model and change the input size in train.py?

and if i don't have much 1920 x 1080 sized samples , is that better to let the input image resize to like 640 x 480 to do the inference
rather than resizing the train image to bigger size ?