MSCV Student Projects

2026 Student Project Sites

3D Asset Generation: Semantic Part Decomposition

Multi-Modal Fusion and Synthetic Data Generation for Adverse Driving Scenarios

3D Asset Generation: Generating Action Primitives

Scene-Aware, User-Guided Generation of 3D Human-Object Interactions

Grounded Vision-Language Understanding for Warehouse Automation

Vision-Based Runway Pose Estimation for Autonomous Landing

Slot4D: Learning Object Slots for 4D Reconstruction

Learning 3D-VLAs with Noisy Miscalibrated Data

Scaling Vision–Language–Action Models for Robotic Manipulation through Simulation

Rasterized Representations for Autonomous Driving

Disentangled Gaussian Transformers for Animatable Humans | Cooperative Foundation Models For Object Detection Active Learning

Monocular Vision for Obstacle Detection in Autonomous Aircraft Operations

Situational Safety for Embodied Agents

Bioreactor Digital Twin

FlowGS: Optical Flow-Guided Gaussian Deformation for Streaming 4D Reconstruction

3D Scene Understanding

Goal Conditioning for Manipulation Tasks

Joint 4D Human and Deformable Object Capture Studio

Pitcher Tell Detection

Multimodal Medical Triaging on UAV Data for DARPA Triage Challenge

2025 Student Project Sites

Photometric Image Enhancement For Robotic Endoscopy

4D reconstruction From Monocular Videos

LVLM Multi-image VQA Dehallucination

Adaptive Data Collection For High Fidelity Gaussian Splatting

World-Grounded Human Mesh Recovery from Video

OnlineHMR: Video-based online world-coordinates human mesh recovery

Text-Driven Motion Generation for Hand-Object Interaction

Temporally Hierarchical Scene Graph Generation for Video Question Answering

Breaking Depth Estimation Models with Semantic Adversarial Attacks

Dual-Path Diffusion Sampling for Training-Free Video Editing

Flow3r: Factored Flow Prediction for Visual Geometry Learning

Debiasing vision datasets

2025 Group 13 Project (P13)

Medical Segmentation with Foundation Models: A Prompt-Based, Text-Guided, Training-Free Pipeline

System Identification from Multi-View Videos

Finding Dents and Dings With a Drone using Structured Light

Computer Vision for Cinematographic Motion Control

Improving Robot Manipulation with 3D Vision Models

Scalable Video Creating and Editing

Making Video Foundation Models Practical: From Physical Modalities to Fast Inference

3D Gaussian Human Enhancement

2025 Team 16-1 Project

End-to-End Infrared UAV Mapping: An Integrated Framework for Thermal Scene Reconstruction

Reinforcement Learning for Noise Steering in Diffusion-Based Driving Models

2024 Student Project Sites

Automatic Hitting Angle Estimation

Generating Real-time Egocentric Viewpoint Scenes

Dynamic Reconstruction of Non-rigid Scenes from Monocular RGB-D Videos

Large Scale Camera Array Calibration via SfM

Learning novel tool use

Why is Sparse-View 4D Reconstruction Hard?

Learning diffusion priors for video amodal segmentation

Alignment for Vision-Language Foundation Models

Driving in Bad Weather

MS Computer Vision Capstone | Team 10

3D Face Fitting and Attribute Classifier Debiasing

Using compressed video cues for faster video transformers

Towards Universal 3D Lifting

Sparse view 3D Pose and Reconstruction

Towards Universal State Estimation and Reconstruction in the Wild

Multimodal Visual Learning for In-Bed Human Mesh

Scene 3D Reconstruction via Diffusion Distillation

Sparse-view Pose Estimation via Analysis-by-synthesis

2023 Student Project Sites

Sensor Simulation with Dynamic Objects

FairToken: Learning Fair Text Representations

Autonomous Driving in Adverse Weather Conditions

Neural Radiance Imaging

Uncertainty in Image classification

2022 Student Project Sites

Open-world Panoptic LiDAR Segmentation

Neural Re-Photography

Automatic bias removal from datasets

Neural Synthetic Wavelength Interferometry

Omnidirectional Multi-view Stereo on Edge Devices

Depth Estimation in Low-light Environments for Autonomous Navigation

One Million Generative Models

Platform Pittsburgh, Smart Cities

Dynamic Implicit Neural Representations for Avatar Animation

Modeling and Understanding Pedestrian Behavior

Multi-Camera Pose Estimation

Automate Pose Estimation for Robotics Manipulation

Unsupervised Online Human Detection and Tracking

3D Face Reconstruction using Mobile Device

Sparse-view 3D Reconstruction

2021 Student Project Sites

Reinforcement Learning for Self Driving Cars

Autoregressive Conditional Generation using Transformers

Dynamic Octrees for Efficient Volumetric Modeling

Forecasting for Autonomous Driving

Panoptic Lidar Segmentation for Autonomous Driving

Multimodal Fusion for Autonomous Driving

Millimeter-wave Radar SLAM

Graph Neural Networks for Vector Map Automation

Human Pose Estimation and Tracking

Neural Rendering of Faces

2020 Student Project Sites

Long Term Mapping for SLAM in Dynamic Environments

Semantic Map Representation

Object Detection in Infrared Images

Reinforcement Learning for Self Driving Cars

Self-supervised 3D Object Detection and Shape Completion

Multimodal 3D Object Detection

Updating HD Maps with Videos from Vehicles

Learning Implicit 3D Representation for Arbitrary Shapes

Consistent Multiview Color Correction

Dense Keypoint Detection on Hands

Self-Supervised Visual Concept Learning

Semantic Facial Image Manipulation using 2D/3D modalities

Real-time Photorealistic VR Facial Animation

Image-based Localization for Autonomous Drones

Self-supervised Representation Learning for Deformable Objects

2019 Student Project Sites

SLAM in Dynamic Environments

Online Calibration

Tracking Two-Hand Interactions

Realistic Human Modeling for Virtual Reality

Automatic Data Augmentation

Deep Slope Estimation with Formal Verification & Cloth Parts Detection

Deep Prediction for Self-Driving Cars

Sony Chef – Learning a Simulator for Cooking

Embedded Pedestrian Tracking and Detection

Sensor Fusion with Single Photon Detectors

RGB Guided Depth Map Super-resolution

Visual Inspection for Aircraft & Power Lines

Deep 3D Mesh Reconstruction

Automatic Photo Composition

Open Vision Platform for Smart City Intersections & VQA from Camera Networks

2018 Student Project Sites

Model Compression

Exploiting Synthetic Data for Street Scene Segmentation

Platform Pittsburgh

Adversarial Learning for Diabetic Retinopathy

Generating risk map for Aerial Drones landing in Urban area

Learning Person-Specific 3D Face Reconstruction with Quantitative Analysis

Target-relevance Enforcement

Indoor scene understanding using long-term time-lapse imagery

Photorealistic 3D Hair Reconstruction

Semi Automatic Point Cloud Annotation

Semi-supervised Similarity Learning On Product Facings

3D detection in Large Scale Point Cloud Data

3D Vision and Learning to predict deformations from CAD models