profile image

Oluwafemi Lamusa

AI Engineer

3D | Computer Vision | AR/VR | Robotics

Contact Me

About Me

A visionary in Computer Vision with extensive expertise and a profound passion for Computer Vision, Deep Learning, 3D Design and Robotics. I have hands-on experience in Camera Calibration, Visual Odometry, SLAM, 3D Reconstruction, 3D Data/Point Cloud Processing, CNNs, VAEs, and Stable Diffusion Models.

I am committed to self-directed learning and personal development, continually building upon past achievements to drive progress and excellence. I thrive on generating innovative ideas and addressing challenges with creativity and originality.

Projects


project name

3D Surface Reconstruction Using Nvidia NKSR

This project showcases my entry for the NVIDIA NKSR Sweepstakes competition. The objective of the competition was to create a 3D model from captured point cloud data, utilizing NVIDIA's NKSR technology.

Find out more

Github Repo

project name

2D Pose Estimation From Images/Videos

Pose estimation and 3D human mesh recovery from a single RGB image, or video using existing technologies such as HMR, HMR2, SMPL, OpenPose, I2LMesh.

Github Repo

project name

License Plate Number Recognition

Automatic license plate number recognition from a video stream. This project identifies various vehicles from a traffic video and tracks them and their corresponding license plate numbers in real time.

Find out more

project name

Point-cloud Processing, Augmentation, Semantic Segmentation

Processing, augmentation, and semantic segmentation of 3D point-cloud data from drone-captured aerial images.

Find out more

project name

Computational Photography and Image Processing

Computational photography operations and image processing using OpenCV and custom filters.

Github Repo

Demo

project name

Camera Calibration & Pose Estimation

Camera calibration is performed to determine a camera's intrinsic parameters (focal length, optical center, and lens distortion), as well as the extrinsic parameters (translation vector and rotation matrix) using OpevCV library. This enables accurate 3D measurements and computer vision applications.

Github Repo

project name

CAD/Mechanical Design

CAD and mechanical projects involving designing of 2D/3D models, assemblies, and simulations using various tools: AutoCAD for detailed mechanical drawings, Inventor for parametric 3D modeling and automation, Fusion 360 for integrated CAD/CAM workflows, SolidWorks for complex product design and engineering analysis, and Plant 3D for piping and industrial plant layouts.

View Samples

project name

Fashion/Clothing Design

Fashion and 3D garment design projects involve creating realistic clothing, accessories, and fabric simulations using CLO3D and Marvelous Designer for dynamic garment draping, pattern-making, and animation, while Blender is used for detailed modeling, texturing, and rendering of fashion accessories and digital assets.

View Samples

project name

Product Rendering

Product rendering projects involve creating high-quality, photorealistic visuals using Blender and 3ds Max for 3D modeling, texturing, lighting, and rendering, bringing product concepts to life for marketing, presentations, and design validation.

View Samples

project name

Architectural Design

Architectural rendering projects using Revit for precise 3D building modeling and documentation, while Blender enhances designs with realistic materials, lighting, and high-quality visualizations for presentations and client approvals.

View Samples

Work Experience

Computer Vision Engineer - Freelance (2022 - Present)

I have leveraged advanced techniques and technologies to solve complex visual data processing challenges. My experience includes utilizing OpenCV for camera calibration, developing depth estimation models using stereo vision, and applying 3D deep learning methods to estimate human pose and shape from images. I have researched and implemented cutting-edge visual odometry algorithms, and used tools like SMLP, HMR, and STAR to create 3D mesh representations of human bodies. Additionally, I have worked extensively with Open3D for 3D data processing, performed data augmentation on 3D point clouds, and executed 3D reconstruction using techniques like photogrammetry, 3D gaussian splatting and NeRF. My work also includes processing images into point cloud datasets, as well as annotating and labeling these datasets for classification and segmentation purposes.

3D Machine Learning Engineer - Elamr Studios (2021 - 2022)

I developed and implemented machine learning models for predictive analytics and classification, with a focus on optimizing deep learning models for tasks like object recognition, segmentation, and reconstruction. My work involved utilizing deep learning frameworks to build and train CNNs, GANs, and stable diffusion models, as well as processing and pre-processing 3D data from various sources, including point clouds and voxel grids. I also specialized in feature selection, feature engineering, and hyperparameter tuning, creating models that incorporated polynomial/basis functions and interaction features. Additionally, I built classification and predictive models to analyze large datasets, ensuring accuracy by minimizing errors during analysis. To efficiently manage large-scale 3D datasets, I developed custom data pipelines and workflows.

AR/VR Developer | 3D Designer - Elamr Studios (2017 - 2021)

I worked as an AR/VR Developer and 3D Designer, where I focused on creating and optimizing immersive AR/VR experiences. My role involved developing and maintaining AR/VR applications using tools like Unity3D, Unreal Engine, Oculus, ARKit, and ARCore, with a strong emphasis on user experience and performance. I performed feature extraction and engineering on 3D data, utilizing techniques such as voxelization and point cloud transformations. Additionally, I created 3D models, animations, and interactive elements in various formats (FBX, USDZ, GLB, GLTF, OBJ) to enhance these experiences. I also designed detailed 3D models and animations for product visualizations, marketing, and promotional materials using Blender, and utilized CAD software like SolidWorks, AutoCAD, and Autodesk Inventor for technical drawings and specifications.