Main Article Content

Abstract

Activity analysis from videos is a highly active area of research with applications over a wide spectrum including health monitoring, intelligent surveillance systems, human computer interaction, robot learning, sports analysisand context based retrieval. As with other research areas of computer vision, recent work has focused on applying deep network models to accomplish the task of activity recognition as it dramatically improves the recognition performance by exploiting the video frame structure. The survey basically captures the current perspective of deep model based solutions for activity recognition from video sequences and imparts potential directions for future research. It also provides comparative view of video datasets usedin recent literature.

Article Details