Groundlight Introduces AI-Powered Counting Mode for Precise Object Counting Without Machine Learning Expertise

SEATTLE -- (BUSINESS WIRE) -- Groundlight has launched its new "Counting Mode," which enables developers to effortlessly and accurately count objects in images using a blend of advanced vision-language models, swift edge inference, and continuous live human oversight. This feature aims to democratize computer vision by allowing businesses to get precise counts without requiring a team of machine learning experts. Many Video Management Systems (VMS) currently offer built-in analytics, such as counting people or vehicles, but these are often based on pre-trained models that are rigid and not tailored to specific business needs. These pre-existing features are generally designed for common scenarios and fall short when it comes to the unique requirements of individual organizations. For instance, a retail store might need to differentiate between the number of shoppers and employees, or trigger alerts when someone enters with a roller bag suitcase. Such specialized tasks are beyond the capabilities of most off-the-shelf VMS solutions, leaving businesses with the choice of either settling for insufficient data or investing in expensive custom solutions. With Groundlight’s Counting Mode, businesses of various sizes can achieve instant and precise object counting using simple natural language commands. The deployment process is streamlined, and the machine learning model can typically be up and running within a couple of days. Users simply describe what they want to count, and Groundlight's system takes care of the rest, analyzing images from existing cameras and delivering rapid, accurate results. The ease of integration, real-time insights, and high accuracy make it a valuable tool for automating counting tasks across different industries. According to Leo Dirac, CTO at Groundlight AI, "Our goal is to make computer vision as easy as asking a question. With Counting Mode, businesses can quickly and accurately count objects in their environment, whether it's parts on an assembly line or products on a retail shelf, without needing a team of machine learning scientists." For a practical example, consider a retail store that wants to gather accurate footfall traffic data but exclude employees wearing company uniforms. Groundlight’s system can count both groups separately, ensuring precise and usable data. Developers seeking to implement this feature can find a step-by-step guide in the GitHub repository linked below. Additionally, if the system encounters any ambiguity regarding user requirements or image content, it escalates the task to Groundlight’s Cloud Labelers, real humans who verify and correct the data in real-time. A detailed explanation of this process is available in a recent webinar recording featuring members of the Groundlight science team. Key Features of Groundlight AI’s Counting Mode: Natural Language Input: Users can specify the objects they want to count in plain language. Rapid Deployment: The machine learning model can be operational in just a couple of days. Seamless API Integration: Easy to incorporate into existing systems and workflows. Real-Time Analytics: Provides immediate object counts from images and video streams. Human Verification: Ensures reliability through continuous live oversight. Enterprise-Grade Accuracy: High precision suitable for critical business applications. Counting Mode is now available to Groundlight users with a Business-level account, and detailed documentation on creating a counting application is provided in the Python SDK Guide. Whether you're enhancing security, optimizing manufacturing, or gaining custom retail analytics, Groundlight’s Counting Mode offers a fast and intuitive solution for obtaining object counts from images and streams. To learn more, visit: code.groundlight.ai