This model takes a squished image as input and predicts whether its width or height has been squished, along with the squish ratio. For example, if an image is resized from (w, h) = (256, 256) to (128, 256), the model output will be (0.5, 1).
This model takes a squished image as input and predicts whether its width or height has been squished, along with the squish ratio. For example, if an image is resized from (w, h) = (256, 256) to (128, 256), the model output will be (0.5, 1).