This work was done over a CCTV system which captures consecutive HD images at a maximum frame rate of 12 frames per second and stored in a cloud-based server.

The process of generating time-lapse video sequences can be divided into three key stages. In the first stage, the FAST Super-resolution processing is used to provide an environment that will permit efficient frame processing. The second stage is where each frame is analysed and classified according to three factors: day/night detection, detection of blurred images and activity detection. The third stage discusses both timestamp removal and temporal smoothing to ensure smooth transition between frames.

The front-end for this project was done in Vue.js/Nuxt.js and the back-end was an API done in Elixir that will run a python script which will do the image processing step.