Research Projects
Exploring robotics, computer vision, and autonomous systems through hands-on research at Northeastern University and Mercedes-Benz R&D.
Vision Foundation Models for Engineering Document Intelligence and Manufacturing Inspection: A Survey
A PRISMA-informed survey of 94 studies on foundation models for engineering drawing intelligence and manufacturing inspection, identifying domain-shift and evaluation gaps across SAM, CLIP, DINOv2, and multimodal models.
One-shot Imitation Learning via Interaction Warping
A method for teaching robots to manipulate novel objects from a single demonstration by decoupling learned actions from object geometry.
3D Object Detection
LiDAR-based 3D object detection using voxelization and bird's eye view for autonomous driving applications.
LEGO LOAM vs SC-LEGO LOAM
Comparative study on LiDAR odometry and mapping using data from NUANCE autonomous vehicle platform.
Image Captioning via Transformers
Transformer-based image captioning model combining computer vision and NLP for descriptive caption generation.
Style Transfer + Super Resolution
Combined style transfer with SRGAN for high-resolution artistic image generation.
Traffic Sign Recognition
Deep learning approach using transfer learning and data augmentation for robust traffic sign classification.
Shadow Removal via Entropy Minimization
Single-image shadow removal using entropy minimization without camera calibration or multiple exposures.