Filtering by
- All Subjects: deep learning
- Creators: Computer Science and Engineering Program
- Member of: Barrett, The Honors College Thesis/Creative Project Collection
The NBA yields billions of dollars each year and serves as a pastime and hobby for millions of Americans. However, many people do not have the time to watch several 2-hour games every week, especially when only a fraction of the game is actually exciting footage. The goal of Sports Summary is to take the ``fluff'' out of these games and create a distilled summary that includes only the most exciting and relevant events. The Sports Summary model records visual and auditory data, camera angles, and game clock readings and correlates it with the game's play-by-play data. On average, a game of more than 2 hours long is shortened to a summary of less than 20 minutes. This summary is then uploaded to the Sports Summary website, where users can filter by the type of event, giving more autonomy and a more comprehensive viewing experience than highlight reels. Additionally, the website allows for users to submit footage they would like to watch for processing and later viewing. Sports Summary creates an enjoyable and accessible way to watch games.
algorithms for recommending new music to users. However, at the
core of their recommendations is the collaborative filtering algorithm,
which recommends music based on what other people with similar
tastes have listened to [1]. While this can produce highly relevant
content recommendations, it tends to promote only popular content
[2]. The popularity bias inherent in collaborative-filtering based
systems can overlook music that fits a user’s taste, simply because
nobody else is listening to it. One possible solution to this problem is
to recommend music based on features of the music itself, and
recommend songs which have similar features. Here, a method for
extracting high-level features representing the mood of a song is
presented, with the aim of tailoring music recommendations to an
individual's mood, and providing music recommendations with
diversity in popularity.
Planning coordination between robots in a multi-agent system requires each robot to know the position of the other robots. To address this, the localization server tracked visual fiducial markers attached to the robots and relayed their pose to every robot at a rate of 20Hz using the MQTT communication protocol. The robots used this data to inform a potential fields path planning algorithm and navigate to their target position.
This project was unable to address all of the challenges facing true distributed multi-agent coordination and needed to make concessions in order to meet deadlines. Further research would focus on shoring up these deficiencies and developing a more robust system.
This thesis is part of a collaboration between ASU’s Interactive Robotics Laboratory and NASA’s Jet Propulsion Laboratory. In this thesis, the training pipeline from Sharma’s paper “Pose Estimation for Non-Cooperative Spacecraft Rendezvous Using Convolutional Neural Networks” was modified to perform pose estimation on a complex object - specifically, a segment of a hollow truss. After initial attempts to replicate the architecture used in the paper and train solely on synthetic images, a combination of synthetic dataset generation and transfer learning on an ImageNet-pretrained AlexNet model was implemented to mitigate the difficulty of gathering large amounts of real-world data. Experimentation with pose estimation accuracy and hyperparameters of the model resulted in gradual test accuracy improvement, and future work is suggested to improve pose estimation for complex objects with some form of rotational symmetry.