Gunnar Hovden Email & Phone Number

Sunnyvale, California, Us

Developing computer vision algorithms for semantic understanding of environments based on data from 3D scanners. Separate 3D models into floors and individual rooms, classify rooms (kitchen, living room, dining room, bedroom, etc.), further separate rooms into walls, floors, ceilings and objects. Calculate room dimensions, area and volume. Change objects and add new objects in models using 3D and panoramic RGB data. Colorize point clouds. Export point clouds to various formats used in AEC (Architecture, Engineering and Construction) including XYZ and Leica Cyclone format (PTS/PTX). Create floor plans, reflected ceiling plans and fenestration plans for AEC. Algorithms for automatic creation of image galleries showing selection of best views for each major area in a model. Blur faces in panoramic images for privacy. Automatic creation of navigation points for Virtual Reality. Mentor new engineers and interns.

Developing computer vision algorithms for Advanced Driver Assistance Systems (ADAS) using a wide variety of cameras and sensors.

Sunnyvale, Ca, Us

Developed video analytic algorithms for the S7000 series of software configurable processors, including face detection and content based optimization of video compression.

Sunnyvale, Ca, Us

Explored computer vision algorithms for use in future consumer products. Developed and tested algorithms, and worked with other team members to incorporate those algorithms into test environments. Worked on areas including color and texture templates, color segmentation, motion adaptive flow, and object recognition.

Sunnyvale, Ca, Us

Developed algorithms and software for the S5000, S6000 and S7000 families of Stretch software configurable microprocessors for processing and recording real-time data from cameras. A broad range of tasks, all optimized for speed, that include:· Video analytics (motion detection, night detection, blind detection) · Video preprocessing (de-interlacing, noise-filtering, scaling, on-screen-display) · Video format conversion· Scene-change detection· Designing in-house fonts· BDF encoding and decoding· JPEG encoding and decoding · G.726 audio encoding and decoding

· Developed a system for classification of user-uploaded images from still image cameras. The system uses neural networks, segmentation, object recognition and color templates. The classification accuracy was 2.5 times higher than existing systems and is used by ImageStation to analyze two to three millions images every month.· Invented, developed and patented a Texture Feature Detector that are used in the SONY DSC-G1 digital camera as part of a Content Based Image Retrieval (CBIR) system. · Invented, developed and patented a Color Feature Detector that are used in the SONY DSC-G1 digital camera as part of a Content Based Image Retrieval (CBIR) system. Increased the return accuracy in tests from 85% (using the cutting-edge technology in industry) to 95%.· Developed distance metric for comparison of faces to produce CMC graphs.· Used structured tensor fields and tensor voting for extracting shapes from natural images.· Developed skin detector.· Developed methods for text detection.· Auto white balance of user uploaded pictures

Ph.D., Computer Engneering

Ms, Computer Engineering

Cand. Mag., Computer Science

Høyskoleingeniør, Industrial Automation

Naturfag