Jiang, HuiSanu, Joseph2020-11-132020-11-132020-042020-11-13http://hdl.handle.net/10315/37859We propose a new unified approach to generate high quality temporal action proposals from untrimmed videos called Incorporated temporal action proposal generation. The concept behind this model is to consolidate the processes that classify the small temporal segments and evaluate the larger proposal features. In doing so, we seek to reduce the total number of computational units necessary for end-to-end proposal generation. For our research, we have conducted our experiments on the action proposal task in the ActivityNet challenge where the goal is to produce a set of candidate temporal segments that are likely to contain a human action. In addendum to the aforementioned research, we also propose an extension to this work by applying video level classification on our proposals. For this work, we emphasize the importance of accurate sequential modelling of temporal segments to properly distinguish between macro and micro-level actions within untrimmed videos and we compare our results with other state-of-the-art spatio-temporal models.Author owns copyright, except where explicitly noted. Please contact the author directly with licensing requests.Computer scienceElectronic Thesis or Dissertation2020-11-13End-to-end proposal generationSequential modelling