• Home
  • News
  • EvaStudy: ACES Students Create Emotion-Aware AI...

News Story

EvaStudy: ACES Students Create Emotion-Aware AI Tool to Transform Studying

What if your study app could recognize when you’re stressed, distracted, or in need of encouragement? That’s the vision behind EvaStudy, an AI-powered productivity tool developed by students Samragyee Dhakal, Alvia Naqvi, and Juhi Chitkara to make studying more focused, emotionally supportive, and human-centered. Samragyee and Alvia, both rising sophomores in the ACES program, are majoring in computer engineering and computer science, respectively.

“EvaStudy is designed to help students stay focused and emotionally supported,” Samragyee explained. “We combined computer vision, emotion detection, and AI-generated feedback to create a smarter study environment.”

The idea emerged from the students’ own experiences as computer science and computer engineering majors constantly struggling to stay on task.

“We saw people filming themselves studying to stay accountable,” Samragyee said. “That inspired us to create a tool that could recognize distraction and emotional fatigue in real time.”

For the team, the core problem was clear: focus. Smartphones, isolation, and academic pressure can all chip away at productivity. EvaStudy responds by not only detecting distractions but by offering timely, human-like encouragement, bridging the gap between productivity and mental health.


Development

What began as a simple tool for blocking distractions quickly evolved into a full-fledged emotionally-aware study companion.

“It started with phone detection,” said Alvia. “Then we added emotion recognition, a chatbot, and session recordings, all to support students more holistically.”

To bring EvaStudy to life, the team built the frontend using Flask, HTML, CSS, and JavaScript, while the backend integrated OpenCV for webcam input and TensorFlow/Keras for emotion classification. They trained a convolutional neural network on the FER-2013 dataset to recognize seven emotional states and used YOLOv5s via Torch Hub to detect distractions like cell phones. Gemini, Google’s multimodal large language model, powered real-time chatbot responses, while OpenCV recorded study sessions and compiled them into 4× speed MP4s. A MySQL database managed user data and session logs, with hashed passwords ensuring login security.

But the project didn’t come together without challenges. The trio juggled complex integrations, a tight timeline, and personal hurdles, including Samragyee falling ill halfway through development.

“One big lesson was learning how to think collaboratively,” she said. “I had to see my part not in isolation, but as part of a system. Like gears in a clock, everything needed to sync.”

Alvia echoed the learning curve: “My initial backend design didn’t work with our browser-based webcam feed. I had to rebuild it all to stream smoothly in real time. Balancing real-time processing for object detection, emotion tracking, and chatbot responses without crashing the system? That was one of the toughest challenges.”

The team overcame these challenges through clear role division and continuous GitHub coordination. Samragyee led database and authentication development, Alvia tackled AI integration, and Juhi focused on crafting a user-friendly frontend.


Key Features

One of EvaStudy’s most distinctive aspects is its ability to detect and respond to both distraction and emotional fatigue in real time. The system uses object detection to identify phones within the video stream, relying on YOLOv5s to draw bounding boxes and assess confidence scores. If a phone is detected continuously, EvaStudy recognizes it as a distraction and prompts the user with a gentle reminder to return to their task.

Emotion detection is just as sophisticated. The system first locates the user’s face in each frame using OpenCV’s Haar Cascade Classifier. These facial regions are converted to grayscale and resized to match the input dimensions of the CNN model, which then classifies the expression into one of seven emotions. EvaStudy tracks recent emotional states using a rolling deque, and if it detects a sustained pattern of negative feelings, such as anger or sadness, it sends an empathetic message, powered by the Gemini API, to boost morale or recommend a break.

What sets EvaStudy apart from traditional productivity tools is its adaptive, emotionally intelligent approach. Instead of relying on rigid time intervals like the Pomodoro technique, EvaStudy adjusts feedback based on how the user is actually feeling in the moment. By blending emotion recognition with context-aware AI, it helps prevent burnout and encourages healthier, more sustainable study habits.


Vision for the Future

The students aren’t done innovating. Samragyee hopes to add multi-factor authentication and stronger password requirements, especially as personalization features grow. Meanwhile, Alvia envisions a future where users can create custom study companions, complete with personalities and voices.

“Imagine having a virtual buddy who cheers you on in your preferred voice,” she said. “It could help reduce the isolation of solo study time.”

Technically, they also aim to refactor the system for production environments, with plans to deploy on AWS, a challenge Alvia is confident in taking on, given her cloud certification.


Impact and Inspiration

Beyond features, EvaStudy points to a larger shift in digital learning: toward emotionally intelligent, human-centered tools.

“I think this is the future,” said Samragyee. “Students deserve tools that understand not just how they work, but how they feel.”

Alvia added: “Many platforms focus on productivity. But emotional well-being is just as important to learning. I hope EvaStudy helps redefine what support in the digital classroom can look like.”

The project has already shaped their future paths. Samragyee discovered a new appreciation for systems thinking and collaboration. Alvia, currently interning at Yellow.ai, has been inspired to explore new applications for LLMs in real-world tools.


Words of Advice

For students dreaming up their own tech solutions, both have words of wisdom.

“Let yourself be bad at things,” said Samragyee. “Learning is messy. But the belief that you can figure it out—that’s what makes building something like this worthwhile.”

Alvia echoed this sentiment, but emphasized the importance of empathy in problem-solving. “Don’t just solve a problem. Solve it in a way that cares about the person facing it,” she said. “When your goal is to support others meaningfully, your solutions become more impactful, and more human.”


Published July 7, 2025