Sandra Sajeev

Title: Sandra Sajeev: Innovator in Temporal Action Localization

Introduction

Sandra Sajeev is a prominent inventor based in Seattle, WA (US). She has made significant contributions to the field of video analysis through her innovative patent. Her work focuses on enhancing the capabilities of classifiers in identifying actions within video clips.

Latest Patents

Sandra holds a patent titled "Prior-driven supervision for weakly-supervised temporal action localization." This invention involves a classifier model that is trained for temporal action localization of video clips. The model ingests training video clips that include actions of interest for identification. It identifies action characteristics within the frames of the video clip, corresponding to known action classes. An actionness score is determined for each frame based on these characteristics. Class activation sequence (CAS) scores are then calculated for sequences of frames, which helps in predicting the temporal locations of actions of interest within the video clip. The classifier is trained with a generated training class of action snippets to identify actions at a video level. Sandra's innovative approach has the potential to significantly improve video analysis technologies.

Career Highlights

Sandra is currently employed at Microsoft Technology Licensing, LLC, where she continues to develop her expertise in video analysis and machine learning. Her work has garnered attention for its practical applications in various industries, including entertainment and security.

Collaborations

Some of Sandra's notable coworkers include Gaurav Mittal and Ye Yu, who contribute to her projects and research initiatives.

Conclusion

Sandra Sajeev is a trailblazer in the field of temporal action localization, with her innovative patent showcasing her expertise and dedication to advancing technology. Her contributions are paving the way for future developments in video analysis.