UCF-CAP, Video Captioning in The Wild

Demand for efficient image transmission and storage is increasing rapidly because of the continuing growth of multimedia technology. in this paper, we proposed an image compression method based on the recognition of the importance of regions in images. As not all the information in an image is equally useful, we can identify important regions in an image for high fidelity compression and accept a comparatively more lossy compression about less important regions. in this work, first, we segment images to two parts, namely, foreground and background, where the foreground represents the more important component and the background is of less importance. Then we apply optimal mass transportation mapping in a GAN framework to magnify the foreground and shrink the background while keeping the shape and total image area unchanged. As a result, the ratio of foreground to background is larger than the ratio of the original image. This ratio is controllable in our process. To restore the image, we apply a GAN model to the compressed and recover the ratio of foreground and background using an optimal mass transportation map. Test results show our method is highly effective in reconstructing detail of important components while achieving a high compression ratio.