Then we extracted the encoder that produces the vector representation of the input images and used it as the basis for a system that computes a similarity score (above). We split this data into training and test sets and trained a network to predict which of each pair of reconstructed images human annotators preferred. On average, the annotators spent 56 seconds on each sample. They are asked to pick the image that is closer to the original. Annotators are presented with two versions of the same image reconstructed from different compression methods (both classical and learned codecs), with the original image between them. We call this deep perceptual loss.įirst, we created a compression training set using the two-alternative forced-choice ( 2AFC) methodology.
In other words, to train our image compression model, we used a loss function computed by another neural network. We drew on this observation to create a loss function suitable for training image compression models. The downstream processing normalizes the encoder outputs and computes the distance between them. F is the encoder learned from the image-ranking task. The architecture of the system we use to compute deep perceptual loss.
0 kommentar(er)
