AuthenTech
Bringing Clarity to Education Proofs.
Created on 7th September 2025
•
AuthenTech
Bringing Clarity to Education Proofs.
The problem AuthenTech solves
Certificate Verification
-
Students can confirm their degree/diploma is genuine.
-
Employers can quickly check a candidate’s credentials before hiring.
-
Universities/colleges can verify transferred students’ records
Preventing Fraud
-
Stops fake certificates from being used for jobs or admissions.
-
Helps government or recruiting agencies detect forged documents.
Digital Record Keeping
-
Users can maintain a verified digital copy of their certificates.
-
Institutions can create a secure, searchable database for alumni records.
Instant Certificate Info
-
Extracts details (Name, Roll No, Course, CGPA, Issue Date) with OCR.
-
Saves users from manually typing and sharing details.
Challenges we ran into
Challenge Faced
One of the biggest hurdles we faced was low accuracy in our YOLO model during certificate detection. Initially, the model was misclassifying text regions or failing to detect certain fields because our dataset was small and lacked diversity. Many certificates had different layouts, fonts, and resolutions, which made it difficult for the model to generalize well.
How We Solved
-
To overcome this, we focused on better data acquisition and augmentation.
-
We collected certificates from multiple sources (different universities and formats).
-
We preprocessed the data by resizing, normalizing, and applying augmentations (rotation, brightness changes, noise addition) to simulate real-world scenarios.
-
We also re-labeled a portion of the dataset carefully to ensure high-quality annotations.
After retraining with this improved dataset and fine-tuning YOLO’s hyperparameters (confidence threshold, learning rate), we saw a significant boost in model accuracy and consistent detection of certificate fields.