A Greek innovative start-up ICT company has developed an integrated platform for industrial computer vision applications. The offered software is a core platform of algorithms that is adjusted to each user needs in order to make their cameras understand the physical world around them. The platform is highly customizable to make it fit to each client specific needs, train and fine-tune the algorithms on the specific needs and produce robust, market-ready products. This is an industry-grade software solution based on state-of-the-art computer vision (3D vision and deep learning), photogrammetry and remote sensing knowledge. Τhe algorithms run in real-time, on low-power devices near the production line facilitating inspection and quality control in a great variety of industries.
The platform detects a wide variety of objects, based on the specific tuning on the images provided by the user. The platform can utilize a single camera, or a stereo camera for having direct world dimensions estimation. Moreover, the platform can exploit still images, or video streams, either offline or even online, depending on the needs of the user. The used architecture is scalable and modular, in order to adapt to each customer needs.
Additionally, the software can use any image data from devices such as a video surveillance camera, machine vision camera, or a mobile phone camera. The only prescribed specification is to have the necessary image spatial resolution that is needed to distinguish an object on the image. A dedicated handheld imaging device has also been developed, based on embedded machine vision cameras and a powerful processor on low power consumption. This device allows stereo-imaging and real-time object detections and dimension measuring. It can be handheld or mounted on a robotic vehicle, or even on a car.
The proposed software solutions are ready to integrate to existing operating systems and can be adjusted to any tailored-made user interface to match the user’s market audience. The solutions can be rapidly trained to any new detected object, minimizing the time-to-market. They can also be adapted to a wide variety of cameras and conditions while they are scalable to mass data inputs. They are also scalable from handheld devices to remote servers in real-time while they can be adjusted and augmented to realize new features and satisfy new needs while working aside other sensors’ data (such as inertial measurement unit, global navigation satellite system, depth sensors etc). The software solutions can also exploit the addition of a second camera for real-time 3D information recognizing static or moving objects and tracking through time. They are accurate and they can detect not only objects and people, but also incidents.
The company seeks to find applications from industry which need to harvest the power of images for automating and accelerating their production pipeline. The company is also interested in cooperating with partners that are already in the market, mainly ICT companies in order push the object detection software to production and/or integrate it with larger systems. The collaboration sought is commercial agreement with technical assistance provided.