Detecting and Counting Empty Seats in Classrooms and Testing with YOLOvX App
Team Members
• Sai Alim: Led dataset curation and annotation efforts, ensuring high-quality data for training.
• Mishra Siddharath : Managed data augmentation and preprocessing pipelines, improving model robustness.
• Panda Jignesh : Handled deployment and testing on the YOLOvX Android app, focusing on real-time inference. Coordinated project documentation and created the demo video, showcasing real-world application.
• Pappachan Litto : Spearheaded model training and fine-tuning, optimizing hyperparameters for better performance.
Use Case Importance
Detecting and counting empty and occupied seats in classrooms, libraries, auditoriums, and public spaces automates occupancy monitoring and minimizes manual supervision. The system provides real-time seat availability information, helping institutions optimize space utilization, improve student convenience, and support smart campus infrastructure.
Data Collection and Annotation
Data Collection:
Short video clips and real-time images were captured using smartphone cameras inside classrooms and seminar halls under different lighting conditions, camera angles, and seating arrangements. Frames were extracted from the videos using a custom Google Colab notebook, producing a diverse dataset suitable for object detection training.
Annotated Classes:
- Empty Seat
- Occupied Seat
Annotation Tool:
- Makesense.ai
Dataset Split:
- Total Images: 1372
- Training: 70%
- Validation: 20%
- Testing: 10%
Sample Images:
- for Empty Seats:
The dataset includes multiple classroom layouts, varying seat arrangements, partial occlusions, and real-world student occupancy scenarios to improve model robustness and accuracy.
Model Training and Validation
Model & Version:
YOLOv8n (Nano) / YOLOvX lightweight object detection model trained using Ultralytics on Google Colab.
Training Details:
- Epochs: 100
- Batch Size: 16
- Image Size: 640×640
- Learning Rate: 0.01
- Early Stopping Patience: 10
Augmentations Used:
- Flip
- 90 degree Rotate
- Crop
- Rotation
- Shear
- Grayscale
- Hue
- Saturation
- Brightness
- Exposure
- Blur
- Noise
Monitored Metrics:
- mAP@0.5
- Precision
- Recall
- Training Loss
- Validation Loss
Performance Improvement:
Additional frames were extracted from classroom videos to increase dataset diversity and improve detection accuracy under real-time conditions.
Model Deployment and Testing with YOLOvX App
Performance:
Achieved smooth real-time inference on mobile devices with efficient seat detection and occupancy counting.
Deployment:
The trained YOLOv8 model was deployed and tested using the YOLOvX mobile application.
Testing Included:
1.Live camera testing
2.Real-time seat counting
3.Multi-seat occupancy detection
4.Different lighting condition testing
5.Classroom and seminar hall testing
Inference Capability:
The model successfully detected both occupied and empty seats with high accuracy and low latency.
Demo Video :
Real-World Applications
- Smart Classroom Monitoring
- Library Seat Availability Systems
- Auditorium Occupancy Analysis
4.Examination Hall Monitoring
5.Smart Campus Automation
6.Public Waiting Area Management
7.Space Utilization Analytics
Conclusion:
The YOLOv8-based Empty Seat Detection System successfully automates real-time seat occupancy monitoring using computer vision and deep learning techniques. The model demonstrated reliable performance under varying classroom conditions and provided accurate empty and occupied seat detection. This project highlights the effectiveness of lightweight deep learning models for smart infrastructure and real-time monitoring applications.