Final-year project catalog

Find a project you can actually build

Working IoT, AI/ML and embedded builds with abstracts, component lists, documentation, and developer support — ready for your final-year submission.

651 projects live
Reset
FYP-DL-MAIN-029 Medium
Object Tracking in Video using Deep Learning

This project develops object tracking in video using a Transformer-based model (BERT/ViT) trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-015 Medium
Speech-to-Text Transcription using Deep Learning

This project develops speech-to-text transcription using a CNN-LSTM hybrid architecture trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-030 Medium
Deep Learning-Based Currency Note Fraud Detection

This project develops currency note fraud detection using a GAN-based generative architecture trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-014 Medium
Deep Learning-Based Named Entity Recognition System

This project develops named entity recognition system using a GAN-based generative architecture trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-013 Medium
Text Summarization using Transformer Models

This project develops text summarization using a Transformer-based model (BERT/ViT) trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-019 Medium
Chatbot using Sequence-to-Sequence Deep Learning Model

This project develops chatbot using a ResNet/EfficientNet backbone trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-024 Medium
Deep Learning-Based Anomaly Detection in Surveillance Video

This project develops anomaly detection in surveillance video using a 3D-CNN for spatio-temporal data trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-DL-MAIN-011 Hard
GAN-Based Image Super Resolution System

This project develops gan-based image super resolution system using a ResNet/EfficientNet backbone trained on an appropriate image/video/audio/text dataset. The trained network is optimized for inference speed and served through a real-time demo application.

FYP-ML-MINI-047 Easy
Crop Recommendation System based on Soil Data using ML

This project builds a data-driven pipeline for crop recommendation system based on soil data using K-Means and hierarchical clustering trained on historical/tabular data. The final model is wrapped in an interactive dashboard so non-technical users can get predictions and insights instantly.

FYP-ML-MAIN-043 Medium
Traffic Accident Severity Predictor using ML

This project builds a data-driven pipeline for traffic accident severity predictor using Random Forest and XGBoost trained on historical/tabular data. The final model is wrapped in an interactive dashboard so non-technical users can get predictions and insights instantly.

FYP-ML-MINI-046 Easy
Employee Productivity Prediction using ML

This project builds a data-driven pipeline for employee productivity prediction using ensemble stacking of multiple classifiers trained on historical/tabular data. The final model is wrapped in an interactive dashboard so non-technical users can get predictions and insights instantly.

FYP-ML-MINI-048 Easy
Loan Approval Prediction using Ensemble ML

This project builds a data-driven pipeline for loan approval prediction using regression models (Linear, Ridge, Lasso) trained on historical/tabular data. The final model is wrapped in an interactive dashboard so non-technical users can get predictions and insights instantly.