Computer Vision Intern, Sentics GmbH
June 2022 — December 2022
• Developed an algorithm that accurately estimated the base point of an object using pose keypoint data from TRTPose and their 2D-3D correspondence, resulting in a 100% improvement in the accuracy of object location estimation.
• Conducted extensive research and experimentation with various object and keypoint tracking methods and presented detailed findings and recommendations to the team, providing valuable insights and contributing to the project's success.