OpenVINO Reviews

OpenVINO is described by users as an open-source, cost-free solution with no associated licensing fees. The software itself incurs no costs, enabling enterprises to access its full suite of tools and optimizations without payment. While edge devices like cameras may involve additional expenses of around $150, OpenVINO's alignment with similarly free, open-source products such as NVIDIA TensorRT and Microsoft's Onyx makes it a cost-effective choice for businesses.

• "We didn't have to pay for any licensing with Intel OpenVINO. Everything is available on their site and easily downloadable for free."

OpenVINO is primarily used for deploying computer vision models on edge devices and servers with a focus on balancing speed and accuracy. Teams utilize it in applications like retail recognition, machine vision, and surveillance analysis. They test models in environments such as Intel's sandbox and optimize models from libraries like TensorFlow or PyTorch. OpenVINO is deployed on platforms like Azure, Raspberry Pi, and Intel VPU for varied use cases, including video analytics and budget-conscious solutions.

Customers are pleased with OpenVINO's technical support, finding it responsive and thorough. Intel's support is highly rated, with effective communication methods via Discord and email. Although the documentation is detailed and necessitates engineering skills, users manage by leveraging comprehensive resources available online. Feedback from partnered users highlights beneficial support, while independent users appreciate the open-source accessibility and robust documentation. Engineering expertise is important for navigating compatibility challenges and model conversions.

Users found OpenVINO's initial setup to vary in complexity. Some reported difficulties with multiple steps and Docker errors, taking hours to complete. Others highlighted ease due to its integration and pre-configured binaries. Those with technical expertise noted faster setup on Ubuntu, but Windows posed more challenges. Documentation aids the process, but advanced skills are essential, especially for containerization and model tuning. It's not suited for beginners, requiring experienced engineers.

Users find OpenVINO's scalability challenging, especially with multiple containers or streams. Some experience positive scalability on server machines, yet it seems less suited for extensive clusters. Engineers and end-users appreciate its utility with Intel devices, though it's perceived as better for cost-saving with CPUs rather than broad scalability. Feedback suggests scalability might not have been a primary design focus, limiting its effectiveness in larger deployments.

OpenVINO shows strong stability with consistent performance across x86 processors. Users report no bugs or crashes, noting its reliability, although occasional production breakdowns like overload occur. Installation and usage clarity can be improved. It effectively handles traffic analysis with multiple cameras but faces limitations with additional hardware. Updates have positively impacted stability. Intel engineers' familiarity with hardware ensures quality, making OpenVINO efficient on Intel hardware. Users appreciate the improvements despite initial issues.