Reimagining Classrooms with GradeWise
In Thailand's packed classrooms, where teachers juggle up to 50 students at a time, the need for smarter, more efficient teaching tools is clear. That's where GradeWise comes in with its bold aim to transform how teachers approach both grading and teaching, with the help of artificial intelligence (AI).
The story of GradeWise starts with a simple question: How can we make teachers' workloads more manageable? Thai educators spend approximately 84 out of 200 academic days on non-teaching activities like grading, which significantly reduces the time available for lesson planning, student support, and professional development. The team aims to address this challenge by automating the grading process and providing educators with detailed insights into student performance.
The technical framework of GradeWise consists of two primary components: Optical Character Recognition (OCR) and the web-based platform that supports user interaction. The development team uses Tesseract OCR in combination with YOLO (You Only Look Once) object detection to accurately extract and interpret handwritten mathematical equations. The system was trained using over 10,000 annotated datasets, with refinements made through platforms like Roboflow to improve its ability to recognize complex mathematical symbols and differentiate between similar characters, such as the letter 'x' and the multiplication sign. Although challenges were encountered, particularly in recognizing fractions and exponents, continuous iteration has led to significant improvements in accuracy.
Today, their OCR system is performing impressively, with an F1-score of 92.4%, precision at 87.8%, and recall at 97.6%. These numbers might sound technical, but in simple terms, they mean the system is highly reliable when it comes to grading student work, making fewer mistakes and capturing more correct answers.
For GradeWise, the user experience was front and center from day one. Drawing inspiration from platforms like Canvas and Google Classroom, the team created a clean, intuitive dashboard that focuses on what matters most: student performance. Their prototype, built in Figma, went through several rounds of feedback and revision, ensuring it is easy to navigate, even for teachers who aren’t tech-savvy.
In an exclusive interview, the team and their advisor shared their reflections on the project’s journey and its future. While they’ve successfully developed a functional prototype, the big question looms: Will GradeWise evolve into a startup?
“We've grasped the potential of this educational tool, especially in the Thai context,” one team member noted. However, the team remains in an exploratory phase, with one semester left to determine the project's full potential. They’re keen on gathering more user feedback, refining the AI’s grading capabilities, and possibly expanding into subjects beyond mathematics, including essay-based assessments through advanced Natural Language Processing (NLP).
What sets GradeWise apart from existing solutions is its deep integration of AI, not just to expedite grading but to enhance the learning process itself. While other platforms focus on efficiency, GradeWise aims to provide personalized feedback, comprehensive student analytics, and adaptive learning insights.
Looking ahead, the team plans to launch a pilot program with select schools, offering free trials to gather real-world feedback. If it goes well, GradeWise could grow into a full-fledged startup, driving educational reforms not just in Thailand but potentially on a global scale.
Imagine an educational system where a tool fosters better learning environments, reduces teacher burnout, and ultimately helps students thrive. These dedicated freshmen at CMKL University, along with their advisor, are working tirelessly to turn this vision into a reality !
