Explainable AI Intrusion Detection for IoT
Ensemble IDS with SHAP graphs that show why each attack was flagged.
Ask about this project- Research
- Final Year
- CSE / SWE / IIT
Arrhythmia AI — ECG image classification with deep learning
Three CNN models read an ECG image, vote on the result and show it on a clean web dashboard — with patient records and an AI chatbot built in.
we can build one for your topic too!
- Department
- CSE / SWE / IIT
- Type
- Research
- Level
- Final Year
- Timeline
- About 2 weeks
- Delivered
- Code, models, app
About the project
Arrhythmia AI reads ECG images and flags possible heart conditions. Three deep learning models study each ECG and vote on the final result, which appears on a simple web dashboard with patient records and an AI chatbot. It is built as a screening aid — a doctor always confirms the result.
Key features
Three CNN models
ResNet50, VGG16 and EfficientNetB7, fine-tuned with transfer learning.
4 cardiac classes
Abnormal heartbeat (HB), MI, previous MI (PMI) or Normal.
Majority voting
All three models vote, for a more reliable final answer.
Patient management
Save patients and keep a history of every ECG result.
AI chatbot
Explains results in simple words and answers questions.
Live REST API
Models served online on Google Cloud Run.
How it works
- 1Upload ECG image
- 2Resize to 224×224
- 33 CNN models predict
- 4Majority vote
- 5Result on dashboard
Tech stack
- Deep learning
- KerasTransfer learningResNet50VGG16EfficientNetB7
- Training
- Google ColabImageNet weights
- Deployment
- REST APIGoogle Cloud RunHugging Face HubWeb dashboard
What you get with a project like this
- Colab notebooks & full source code
- Trained models, ready to use
- Working web dashboard
- Live API on the cloud
- Report & presentation slides
- Viva prep & one-to-one walkthrough
Your project could be next.
Tell us your topic and deadline. Get a free plan and quote — usually within 30 minutes.
no payment needed to ask!