AI4ALL at the University of Maryland
AI4ALL at the University of Maryland
Embedded Files
Summer 2022 Research Presentations
Summer 2022 Research Presentations
For more pictures of how much fun camp was this summer, click here!
ai4all summer 2022
ai4all summer 2022
A three-week residential summer camp, exposing students to personal growth, education, and hands-on experiences in artificial intelligence presented by faculty, guest lecturers, and University of Maryland students.
Meet our Participants
Project Groups
Project Groups
Slightly altered images that result in inaccurate and highly confident predictions are known as adversarial attacks. For this project, we trained a model to classify images from a particular dataset. In order to improve models like this, so that they can properly recognize the images, adversarial training is used. This involves slightly altering images through making changes to pixel values by small, seemingly unpatterned amounts.
This project is primarily about exploring navigation for robots and solutions to the lack of understanding they have about their environment, to help robots achieve Robotic Semantic Navigation (RSN). We attempted this using Object Detection - the process of processing an image and then localizing and classifying all of the objects within that image - and simulated our algorithms first using a Robotic Operating System (ROS) before putting actual robots into action.
For this project, we worked on constructing a machine learning algorithm to predict the movement of pedestrians in a crowd so as to develop clear and direct paths for robots to move through a crowded environment. We made use of data obtained from recordings of people entering and exiting a university and hotel to create our neural network and predict potential motion, etc.
The purpose of this project is to identify which of the eight main emotions (neutral, calm, happy, sad, angry, fearful, disgusted, or surprised) a person is experiencing based on their speech. We utilized the RAVDESS dataset, consisting of 60 recordings (each portraying one of the eight emotions listed) per voice actor, spanning 24 professional voice actors (totaling 1,440 samples) to develop and train a machine learning algorithm in order to accomplish our goal.
Page updated
Report abuse