TwentyBN offers software to analyze images or videos where a great variety of objects have to be detected in uncontrolled environments: it is robust against changes in lighting, partial occlusions and deformed objects. The advantages over traditional CV techniques become most obvious in applications like outdoor plant grows inspection, waste sorting or satellite image analysis. For that deep learning techniques are employed, which replace extensive programming by training the system with samples images. TwentyBN supports their customers during their design, prototyping and development phases.
TwentyBN's expertise is to create Deep Learning solutions. This special machine learning technique makes it possible to solve image and video analysis task by letting the computer learn how to solve the problem at hand. This approach spares the specialized computer vision engineer who has to explicitly program an image processing pipeline. Instead with TwentyBN's Deep Learning software an (almost) non-technical user can teach the system by collecting example pictures of a CV task. For example this way an object recognition task can be created much faster and even more robust than common template matching algorithms.
We offer software to solve object recognition tasks. The standard way is to use it as a cloud-based SaaS solution: users can create applications in a web front-end and use it by integrating the TwentyBN API into their client software. If a cloud-based solution is not an option we also deliver our software as a on-premise installation package.
We support our clients first with general guidance to this new Deep Learning approach and second with feasibility studies to their specific tasks at hand. Furthermore we offer to tailor our standard software modules to specific needs. Please contact us if you have problem traditional machine vision software is not capable of. Very likely that our Deep Learning skills give you the answer.