Brown Digital Repository | Item

Full Metadata

Overview

Year:: 2020
Contributor:: Zhou, Zhiyuan (creator); Paradiso, Michael PhD (creator); Boyum, Spencer (creator); Michael Paradiso (advisor); Brown University. Karen T. Romer Undergraduate Teaching and Research Awards (research program)
Genre:: posters
Subject:: Computer vision; Image processing; Kalman filtering--Mathematical models
Extent:: 1 poster

Files

Description

Abstract:: In recent years, object recognition on image has achieved great success. Well-trained CNNs can successfully detect objects in a single image with high accuracy. However, there remain all sorts of unsolved problems in that of video. This work considers two post-processing models based on Inertial Measurement Unit (IMU) to enhance the accuracy of video object recognition on light-weight devices. Videos are rich with temporal information, and IMU is a cheap and accurate way of accessing it. The work combines temporal information of IMU with two post-processing models: 1) Intersection over Union model and 2) Kalman Filter, both of which require small memory and low compute time, making them an ideal choice for light-weight device. A video data set with IMU data is collected and processed using a popular classification CNN You Only Look Once (YOLO). The recognition results are then passed on to the above two models, and the results are compared to show that IMU data applied to the two models can significantly increase recognition accuracy on both models. A detailed description and codebase of this work can found at https://github.com/paulzhou69/object-recognition-imu

Content

Citation

Zhou, Zhiyuan, Paradiso, Michael PhD, and Boyum, Spencer, "Post-Processing of Video Object Recognition Using Inertial Measurement Unit" (2020). Summer Research Symposium. Brown Digital Repository. Brown University Library. https://repository.library.brown.edu/studio/item/bdr:1138470/

Relations

Collection:

Summer Research Symposium

Each year, Brown University showcases the research of its undergraduates at the Summer Research Symposium. More than half of the student-researchers are UTRA recipients, while others receive funding from a variety of Brown-administered and national programs and fellowships and go …
...

Post-Processing of Video Object Recognition Using Inertial Measurement Unit