Deep-neural-approach-to-colorise-monochrome-images and grayscale videos

This project was created for colourisation of monochrome images. This work of mine was inspired from the works done by Zhang et al. [Refer: Richard Zhang, Phillip Isola, Alexei A. Efros.Colorful Image Coloriza-tion.arXiv:1603.08511 [cs.CV]] .But I extended the idea of the project for colourising all sorts of monochrome images along with videos that are typically grayscale Eg. Surveillance camera footages. We have analysed the working of the model for video inputs by analysing frame rate (in per second) for each every frame and tried to draw a best feed line.

Dataset:

We have used the imagenet dataset for training, testing and validating our deep neural model

Result analysis:

Refer to image.png for results obtained in colourising monochrome still images. Refer to video.png for results obtained in colourising grayscale videos frame by frame.