I am using a script that uses a depth ai camera for live object detection. I use a different script to capture the images used to train the model. My problem is that the object detection script is more zoomed in than my image capturing script (therefore not all objects can be detected). I am using yolov8 so I used 640x640 for the preview size. I know that this is causing an issue because when I set it to 1080p the camera uses its full field of view (although the stream display takes up my whole screen so if anyone also has a fix for that too it would be very useful). The problem then is that setting preview size to 1080p messes up the yolo model as it needs 640x640. Is there any way to send a 640x640 stream to my yolo model without cropping the fov?
I am pretty new to all this and I am working with some pre made code by one of my out of office colleagues so cut me some slack if this question has an obvious answer haha.