Executed comprehensive vehicle model classification in collaboration with IAM Metrics Project team by training multiclass classifiers creating a dataset of over 2,000 web-scraped vehicle images
Engineered specialized web scraper to curate images of 9 distinct car models, serving as the project's data source
Leveraged transfer learning techniques for training classifiers, with the ResNet-18 model excelling and delivering a remarkable F1 score of 0.98
Designed and simulated a versatile rover using ROS and Gazebo, enabling autonomous navigation and precise terrain mapping.
Implemented and optimized SLAM algorithms for real-time mapping of complex, uncharted terrains.
Developed an advanced entity tracking algorithm with OpenCV, enhancing the rover's situational awareness and decision-making capabilities.
Spring, 2022
Computer Vision with Drones
Integrated advanced computer vision algorithms into a drone system, employing PyTorch and Python
Enabled the drone to autonomously follow a person using face tracking and respond to hand/body gestures for control
Spring, 2022
Drone control using an EEG headset
Used an EEG headset to control a drone by analyzing blink patterns and EEG raw data.
Spring, 2022
Hate speech detection using machine learning techniques
Trained machine learning algorithms in python to detect hate speech. Algorithms include: Boosting, Neural Networks, Logistic Regression, Random Forest.The highest accuracy was achieved with the deep learning model(99.6%).
Designed and constructed a four-degree-of-freedom robotic arm, applying kinematics and reverse kinematics principles for precise control and motion planning.
Integrated vision-based perception systems for object recognition and grasping small objects.
Programmed preset motions into the app, enabling the robotic arm to execute tasks such as drawing shapes on paper with precision, demonstrating the successful implementation of motion planning algorithms using MATLAB and Simulink