Vision Framework Team's Goal
- Developing most optimized computer vision framework software on Samsung vision processor in the automotive SoC
Major Responsibilities
- Work on porting/optimizing computer vision kernels (most of major kernels supported by OpenCV)
- Work with major categories of computer vision algorithms including:
- Image processing: vector, drawing, histogram, filters, warping, scaling
- Segmentation: thresholding, contour, watershed, edge detection, grapcut
- Feature detection: FAST, STAR, SIFT, SURF, ORB, Harris
- Object detection: boosting, Haar-like, Hog
- Motion: optical flow, tracking
- Classification: SVM, kNN, Random forest
- Custom algorithms from customer's requirement
- Implement OpenVX standard with highly optimized kernels and custom extensions for vision processor
- Implement OpenCL standard on Samsung vision processor
- Implement Samsung's own style computer vision framework (OpenCV like api)
- Develop test plans and write full automated diagnostic tests, unit tests and integration tests
- Analyze customer issues (Vision framework users such as perception module) and provide support
- Develop software requirement and documentation for safety certification (ISO-26262, ASIL-B)
Preferred Qualifications
- Minimum 3 years of experience in relevant field.
- C/C++ programming skills
- Knowledge of computer architecture and DSP principles
- Knowledge of imaging, computer vision and AI algorithms
- Knowledge of software implementation on embedded systems
- Experience of computer algorithm porting, optimization and performance tuning
- Familiarity with optimizing and debugging code on the DSP or embedded platforms
- Experience in intrinsic style programming, understanding assembly code desired
- Knowledge of OpenVX, OpenCV or OpenCL highly desired.
- Experience of development in multicore and multi-processors architecture is highly preferred