New System Improves Machine Recognition of Facial Expressions

Edith Cowan University (ECU) researchers are developing a new method to help machines better recognize human facial expressions and become more emotionally aware.

“As digital systems like virtual assistants and wellness apps interact more with people, it’s crucial they understand human emotions,” said ECU Ph.D. student Sharjeel Tahir.

A More Human Approach to Emotion Recognition

Rather than using single images to train systems to detect emotions, the team—led by ECU senior lecturer and AI expert Dr. Syed Afaq Shah—adopted a more human-like method: presenting sets of related facial expressions to give machines a wider emotional context.

“Just as we don’t judge someone’s emotions from a single glance, our approach analyzes multiple expressions to make more accurate predictions,” Tahir said. “It helps machines better understand emotions, even when faces appear at different angles or in varied lighting.”

Though not applied to physical robots, the research could shape the development of emotionally aware systems in areas like mental health support, customer service, and interactive learning.

Building Machines That Truly Understand Faces

“We’re building a foundation for machines that truly understand faces—not just see them,” Tahir added.

Co-author and Ph.D. student Nima Mirnateghi noted that their method provides rich visual context, boosting emotion recognition accuracy while remaining computationally efficient. The findings were published in the 2024 International Conference on Digital Image Computing: Techniques and Applications (DICTA).

“By training the model with diverse features in an organized set, we found it could recognize emotional patterns much more effectively,” he said.

Guided by Dr. Shah, Tahir is now focusing on developing artificial empathy in AI agents, enabling them to respond appropriately to human emotions.

Meeting the Rising Need for Emotional Support with Intelligent Machines

“There’s a growing demand for emotional support, and emotionally intelligent machines or robots could help meet that need,” Tahir noted.

Mirnateghi added that the research not only advanced AI’s ability to recognize emotions but also encouraged a deeper investigation into how AI models make decisions.

“Our research group is now concentrating on explainable AI in language models, aiming to reveal the complex processes behind how artificial agents recognize and interpret patterns,” he said.

“By increasing transparency, we hope to design AI systems that are easier for humans to understand—narrowing the divide between sophisticated algorithms and human insight. One key question we’re exploring is: what truly makes a machine emotionally intelligent?”

Read the original article on: Techxplore

Read more: Flying Robot Transforms Mid-Air to Land and Travel on Wheels