News November 24, 2023

Darwin Edge Presents Improvements in AI Face Detection Models at TELFOR

Darwin Edge showcased its latest research on enhancing AI face detection models at the Telecommunications Forum (TELFOR) in Belgrade, Serbia. The presented technique reduces processing times and memory usage on edge devices without compromising accuracy, making it ideal for real-time applications in healthcare, public safety, and beyond. The presentation highlighted Darwin Edge's commitment to advancing efficient, scalable AI solutions for edge environments.

A representative from Darwin Edge presenting the research paper titled 'Improving Efficiency of RetinaFace Models by Controlled Branch Pruning' at the TELFOR event, with a projector screen displaying the title slide and the presenter standing beside it.

Darwin Edge took center stage at the Telecommunications Forum TELFOR, held in Belgrade, Serbia, where we presented the findings of our research paper on enchaning the performance of AI RetinaFace detection models. This work was done in collaboration with the Academy of Applied Studies from Serbia, and our sister company Darwin Digital.

Improving Efficiency in AI Face Detection Software

Our research paper, titled “Improving Efficiency of RetinaFace Models by Controlled Branch Pruning”, was well-received by the attendees of the TELFOR event. The paper explores how the RetinaFace model, a multi-level face detection algorithm which uses different branches to detect faces of various sizes, can be optimized by pruning the branches responsible for detecting smaller and medium faces. This approach is particularly effective in scenarios that foucs on detecting large faces , such as close-up face recordings. The results demonstrated significant reductions in processing times, resulting in enhanced performance and reduced memory usage, making the model suitable for use on edge devices – all without compromising the model’s accuracy, nor requiring model retraining.

To gain access to the original paper, please visit this page.

Implications of Our Research Findings

The findings from this research have broad implications, suggesting significant improvements in the efficiency of AI Face Detection models, particularly for mobile and edge devices. This is especially relevant for  industries requiring high-quality, real-time processing of close up face recordings such as healthcare and public safety. This advancement proves our dedication to providing innovative, efficient AI solutions for edge devices, offering reduced latency, energy savings and enhanced security.

About TELFOR

Presenting our research at TELFOR, a key event in the IT and telecommunications field, was a proud moment for us. The Telecommunications Forum (TELFOR) is a an annual conference that gathers global experts, researchers and IT professionals for discussions on emerging technologies, trends, and relevant research and regulations in the fields of telecommunications, information technology, and related disciplines.

About Darwin Edge

At Darwin Edge, we are dedicated to advancing Edge AI technologies and collaborating with our clients to push the boundaries of what’s possible with IoT and AI. Our expertise spans across various applications from computer vision to audio/signal processing, natural language processing, and predictive analytics, serving industries such as healthcare, wellness, fitness, manufacturing, consumer goods, and public safety. We deliver robust, innovative , scalable and cost-effective AI solutions tailored to each specific industry needs.

For more information or to discuss potential collaborations, reach out to Darwin Edge and join us in developing the next ground-breaking digital innovation.

Subscribe to Our Monthly Newsletter

Subscribe to our newsletter to receive the latest news, insights, case studies, and innovations from Darwin Edge directly in your inbox. Stay informed with updates on our cutting-edge Edge AI solutions, upcoming events, and industry trends.