PopularMajor Project
Artificial Intelligence
Face Recognition Attendance System
Camera-based attendance with liveness detection and analytics
- Python
- OpenCV
- FaceNet
- Flask
- +2
₹4,900₹5,999
4.8/301 reviews
Artificial Intelligence · Major Project
Human Activity Recognition from Video
CNN-LSTM and 3D-CNN models that label actions in video clips
3-4 weeks build
Overview
Trains two temporal models on UCF101 action clips and compares them honestly: a CNN-LSTM that encodes frames and reads them as a sequence, and a 3D-CNN that convolves over space and time together. The report covers clip sampling, the cost of each model, and where each fails, and a demo labels uploaded clips.
What makes it stand up
- 01UCF101 loader using the official train-test splits
- 02CNN-LSTM with a pretrained frame encoder
- 033D-CNN baseline with spatio-temporal convolutions
- 04Accuracy, per-class F1 and compute cost compared side by side
- 05Upload demo with top-5 actions per clip
Module breakdown
- 01
Preprocessing
Frame extraction, fixed-length clip sampling and resizing.
- 02
CNN-LSTM
Frame features from a pretrained CNN fed into an LSTM.
- 03
3D-CNN
Spatio-temporal convolution model trained on stacked frames.
- 04
Evaluation
Split-wise metrics, confusion matrix and failure clips.
- 05
Serving
FastAPI inference endpoint with a Streamlit front end.
After this, you will be able to
- Explain the difference between sequence and volumetric video models
- Defend your clip-length and sampling choices
- Demo action labels on a clip the examiner picks
Technology stack
- Python
- PyTorch
- OpenCV
- FastAPI
- Streamlit
- Level
- Major Project
- Domain
- Artificial Intelligence
- Build time
- 3-4 weeks
- Difficulty
You receive
- Complete, commented source code
- 60-90 page project report (IEEE format)
- Editable presentation deck
- Architecture, ER and UML diagrams
- Local setup and deployment guide
- Sample dataset or seed data
- Viva question bank with answers
- One 45-minute walkthrough call
- 30 days of doubt-clearing support