Dynamic Video Stitching & Optimization Expert
Budget: €1,500 – €3,000 EUR
Requirements:
Advanced expertise in computer vision and image processing
(e.g., OpenCV, CUDA, NumPy).
Proficiency in real-time video stitching algorithms and
optimization.
Experience with multi-core CPU parallelization and GPU
acceleration (e.g., NVIDIA CUDA, OpenCL).
Strong skills in Docker containerization, deployment, and
debugging across diverse platforms (e.g., embedded devices and
servers).
Familiarity with VR/360 video streams, wide-angle camera data,
and embedded systems (e.g., RK3399/RK3588).
Responsibilities:
Redesign the current video stitching algorithm to reduce
cropping, enhance clarity, and support real-time performance.
Replace the traditional fixed algorithm with a dynamic, adaptive
solution (e.g., feature-based stitching or AI-assisted methods).
Containerize the solution using Docker, ensuring it can be
deployed on either remote devices (RK3399/RK3588) or the
server, with flexible configuration for both environments.
On sever deployment: Optimize the algorithm to fully
leverage multi-core CPUs, with an optional GPU processing
mode (e.g., CUDA-enabled) selectable via configuration.
On devices deployment: Deliver usable, well-documented
source code (preferably in Python or C++) that integrates
seamlessly with the existing Gstreamer pipeline.
Debug and test the Dockerized application across multiple camera
setups and deployment targets (remote device and server),
ensuring stability and performance (e.g., <80ms stitching
latency).
Advanced expertise in computer vision and image processing
(e.g., OpenCV, CUDA, NumPy).
Proficiency in real-time video stitching algorithms and
optimization.
Experience with multi-core CPU parallelization and GPU
acceleration (e.g., NVIDIA CUDA, OpenCL).
Strong skills in Docker containerization, deployment, and
debugging across diverse platforms (e.g., embedded devices and
servers).
Familiarity with VR/360 video streams, wide-angle camera data,
and embedded systems (e.g., RK3399/RK3588).
Responsibilities:
Redesign the current video stitching algorithm to reduce
cropping, enhance clarity, and support real-time performance.
Replace the traditional fixed algorithm with a dynamic, adaptive
solution (e.g., feature-based stitching or AI-assisted methods).
Containerize the solution using Docker, ensuring it can be
deployed on either remote devices (RK3399/RK3588) or the
server, with flexible configuration for both environments.
On sever deployment: Optimize the algorithm to fully
leverage multi-core CPUs, with an optional GPU processing
mode (e.g., CUDA-enabled) selectable via configuration.
On devices deployment: Deliver usable, well-documented
source code (preferably in Python or C++) that integrates
seamlessly with the existing Gstreamer pipeline.
Debug and test the Dockerized application across multiple camera
setups and deployment targets (remote device and server),
ensuring stability and performance (e.g., <80ms stitching
latency).