Ibsen Giovanni Bazie
Computer Vision & Spatial AI Engineer
I build vision systems across three levels of difficulty: inspection running in real time on a production line, reading and checking documents and labels, and reconstructing 3D space from photographs. Six peer-reviewed publications, first author on three, and a Best Paper Award at EUSPN 2025. Born in Burkina Faso, working from Hanoi.
Tools & Technologies
Featured Works
View all →FOODSCAN
AI food compliance platform, built under AIOBI. Reads a product label with OCR, then runs it through modular checkers for additives, allergens and labelling rules.
Industrial 3D Asset Detection from LiDAR
Automated pipeline for 3D object detection and classification from raw LiDAR point clouds. Handles noisy real-world scan data with occlusion. Generates structured asset registers for digital twin systems.
Akiya3D: 2D Floorplan to 3D Model
End-to-end pipeline converting 2D floorplan images into interactive 3D models. Research lead on a Japan-Vietnam collaboration working on Japanese residential floorplans.
SpatialDemo: Queryable 3D Scene from Photographs
Turns a set of photographs into a 3D Gaussian Splat scene you can ask questions about. 2D detections are fused into 3D positions by multi-view ray casting, then indexed so natural language queries resolve to objects and their spatial relationships.
Industrial Tray Inspection: Colour and Count
Real-time inspection running on a production line at an industrial automation company in Hanoi. Classical computer vision, no neural network, on a Raspberry Pi with a CSI camera, reporting results to a PLC over Modbus TCP.
3D Point Cloud Change Detection
Detects what changed between two LiDAR scans of the same place. Voxel differencing followed by clustering, evaluated on the Toulouse dataset.
Open to AI consulting, freelance work, research collaborations, and talks.