Berkeley Design Technology, Inc. (BDTI) is an engineering services company focused on computer vision and machine learning. BDTI's expertise lies at the intersection of architectures and algorithms--the design and implementation of optimized algorithms for computer vision and deep learning on a wide range of architectures. BDTI's customers include both companies that provide technology for computer vision and machine learning as well as companies that incorporate these technologies in their products. BDTI is dedicated to helping companies successfully create, understand, and use computer vision and machine learning technology.
• Algorithm Development, designing novel image processing, computer vision, and deep learning algorithms to meet clients’ unique requirements.
• Software Development, utilizing deep experience in applications, algorithms, and architectures design and implement highly optimized algorithmic software for computer vision and deep learning. In optimizing software, BDTI uses appropriate algorithm-level modifications, including more compact data types, data re-use, and refactoring. BDTI will also use multi-threading, data parallelism, optimized libraries, compiler intrinsics, and assembly code in creating highly efficient software.
BDTI provides customized services for computer vision and deep learning algorithms and software. Examples of customer projects include:
(The Lenovo Phab 2 Pro smartphone with Google 3D sensing technology.)
(A highly efficient neural network-based pedestrian detection application implemented and optimized for use in a mobile processor.)
Design of an Algorithm for Food Classification
BDTI designed a neural network-based classifier to identify different types of food. This project enabled an industrial equipment manufacturer to assess the feasibility of neural network-based classifiers in its products.
Design and Implementation of Pedestrian Detection
BDTI designed and implemented a highly effiicient neural network pedestrian detection application for a semiconductor IP company. The application is extremely lightweight, requiring minimal processing resources and power.
Implementation and Optimization of 3D Sensing
BDTI optimized Google's 3D sensing technology, enabling it to run in real time on a Qualcomm mobile chipset. The project required substantial algorithm restructuring and recoding in order to use processor resources efficiently.
For more examples of BDTI expertise and experience, please visit https://www.bdti.com//services/computer-vision-and-deep-learning#examples.