Simple vs complex temporal recurrences for video saliency prediction

This paper investigates modifying an existing neural network architecture for static saliency prediction using two types of recurrences that integrate information from the temporal domain. The first modification is the addition of a ConvLSTM within the architecture, while the second is a conceptually simple exponential moving average of an internal convolutional state. We use weights pre-trained on the SALICON dataset and fine-tune our model on DHF1K. Our results show that both modifications achieve state-of-the-art results and produce similar saliency maps. Source code is available at

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Video Saliency Detection MSU Video Saliency Prediction SalEMA SIM 0.571 # 9
CC 0.636 # 9
NSS 1.63 # 9
AUC-J 0.821 # 9
KLDiv 0.647 # 9
FPS 32.97 # 2