Zeus trains an agent that learns to adaptively modify the input video segments to an action classification network . Zeus is capable of answering the query at a user-specified target accuracy using a query optimizer that trains the agent based on an accuracy-aware reward function . Zeus outperforms the state-of-the-art frame-based techniques by up to 1.4x and 3x, respectively. Furthermore, it satisfies the user’s target accuracy across all the queries, at up to 2x higher accuracy, than frames-based methods. Zeus is able to answer the query with up to 3x more accuracy than frame and window-based approaches. Zeus can also be used to answer queries at higher accuracy than those of its own algorithms. It can be used in a short sequence of videos with a large number of videos that have been analyzed by a computer with a deep reinforcement learning agent or a large amount of data to solve the problem of visual recognition and analysis of the user’s workloads, such as a visual search for a visual recognition tool. It is available in the next version of this version of Zeus’ version of the same software, which uses a

Author(s) : Pramod Chunduri, Jaeho Bang, Yao Lu, Joy Arulraj

Links : PDF - Abstract

Code :
Coursera

Keywords : zeus - accuracy - x - based - user -

Leave a Reply

Your email address will not be published. Required fields are marked *