Delasoft, Inc. knows that the future of technology lies in expanding the capabilities of our systems. Our developers are building intelligent programs and exploring the world of deep learning, so that we can design more powerful software solutions. At Delasoft, we innovate in artificial intelligence with the goal of bringing these capabilities to the field of roadway management.
Machine Learning Platform
MACHINE LEARNING PLATFORM
About the Platform
Delasoft’s machine learning platform is focused on asset extraction and video and still footage analysis. Leveraging the Pytorch Library and neural network capability, our developers have been defining new ways that artificial intelligence and deep learning can support transportation specialists.
Our development team uses supervised and unsupervised learning to train the software in new forms of asset extraction types, robust to noise and changes. The platform is trained on the Manual on Uniform Traffic Control Devices (MUTCD) database, and its image recognition capabilities are currently being expanded to cover a variety of other objects, such as guardrails, jersey barriers, and overhead lights.
We’re always working to improve and experiment with our machine learning software, incorporating it with our other platforms and finding new applications for our systems. Our focus is currently on improving image recognition for road sign inventory and extraction of roadside assets.
Asset Extraction — Delasoft’s Asset Extraction and Inventory tool supports your team in maintenance and management of your inventory. An intelligent system reviews still photos taken along the route to keep track of roadside objects. Keep your inventory up to date and cross-check with old and new records to comply with departmental standards. Our machine has been trained on MUTCD standards and can not only recognize roadsigns but inventory them along with vital data, including a still photo of the sign itself, the sign type, and sign content.