Sequence Models for Vision Forecasting
Introduction Vision forecasting is something I’ve been fascinated with so this post serves as a reflection of my understanding. This post will take you through the latest advancements in deep learning techniques used for sequential modelling, in particular with the use case of time-series vision forecasting. I’ll start with the ideas behind staple architectures, like a vanilla RNN, LSTM, and GRU, and then discuss the advancements made by their successors and why/how each new development did better than the previous one....