Image colorization using Scaled-YOLOv4 detector

Image Colorization is the problem of defining colors for grayscale images. Recently many research works have been conducted to propose fully-automatic colorization methods. However, these papers failed in colorizing images with multiple objects accurately. This might be because dealing whole multi-object image as a single input. Following efforts made last few years, this paper aims at studying effect preceding an object detection phase, such that will each individually well full image. After and image, they are fused together reach more accurate colorized In our work, we used detector (Scaled-YOLOv4) than by state art increase quality results. Comparing results literature, it found using Scaled-YOLOv4 increases Peak signal-to-noise ratio (PSNR) 2.6%. Results different extensions compared, png extension got 5.8% better value Learned Perceptual Patch Similarity (LPIPS) metric JPEG.