Senior Computer Vision Engineer

Description

We are looking for software engineers to enable mixed reality experiences by developing state-of-the-art volume capture techniques. Volume capture is an exciting emerging technology that bridges many subfields of computer vision, as well as computer graphics, systems design, image sensors, and film production. The result is a high-fidelity reconstruction of temporally dynamic, non-rigid scenes, which can then be viewed and experienced on VR/AR/MR platforms. Solving these problems requires deep domain knowledge of both classical and bleeding edge algorithms and techniques, and solving them well requires innovation, creativity, and collaboration across domain boundaries.

Culver City, CA

Areas of specific expertise include but are not limited to:

• 3D Reconstruction: Accurate and robust depth map generation, multi-view and multi-modal data synthesis, and watertight surface estimation

• Multi-Sensor Calibration: Intrinsic and extrinsic calibration of multi-sensor, multi-modal systems of cameras and projectors

• Segmentation: Accurate segmentation of foreground objects and background

• Tracking + Scene Flow: High-confidence sparse 2D/3D feature tracking; dense or semi-dense optical/scene flow

• Scene Understanding: Understanding and semantically labelling content in a scene

• 3D Model Compression: Adaptive simplification/decimation of highly dense point clouds/mesh objects; compression of temporal object sequence

Responsibilities:

• Research, design, implement and optimize algorithms related to the broad problem of volume capture

• Collaborate with team members across different specializations

• Work with data from multi-camera, multi-modal capture systems

• Define metrics for system/algorithm performance

• Stay up to date with state-of-the-art research and industry progress

• Document and organize algorithm design and implementation decisions

• Communicate regularly with the team about research advances, algorithm design, etc

• Participate in strategic discussions around data capture and system design: sensor types and specifications, geometric layout of hardware, types of scenes which reconstruct well/poorly, anticipating future applications, etc

Qualifications:

Expert knowledge in at least one area listed above, as well as:

• Fluent in C/C++ (programming and debugging)

• Experience working with OpenCV

• Strong written and verbal communications skills

• Knowledge of parallel computing, OpenCL, CUDA, GPGPU is a plus

• Knowledge of deep learning techniques is a plus

• Knowledge of optimization techniques is a plus

Education:

• Relevant BS and 5+ years of experience

• MS or Ph.D. is preferred

• Well-qualified candidates are encouraged to apply, regardless of education/experience