Image for AI object detection

I’m training an object detection model using the 300 x 300 pixel training images, as instructed. The Oak-D camera will supply images via the color camera to this AI model. My question is, can I feed 1920 x 300 images for detection? I want to use the full width of the camera image. And I only want 300 pixels of the height dimension. Will this work with the AI detection? Or must I break that wide image down into 300 x 300 blocks and feed them all into the AI detector sequentially? This will really slow things down!