TEAM MEMBERS

• Kirti Shahi
• Ranjitha Muniramaiah
• Venkata Naga Bhargavi Puvvadi
• Akash Baidya
• Sindhuja Amudalapadu

PROBLEM DESCRIPTION

• Investigating urgent firearm detection in public spaces like schools and hospitals.
• System notifies authorities upon detection of a displayed firearm, enhancing safety measures.
• The intriguing aspect of this issue lies in its ability to utilize cutting-edge technology like deep learning and live surveillance, thereby improving public safety measures
• For low-level threats, security personnel are notified, while high-level threats involve direct involvement of law enforcement.

ANALYTICS PLAN

• Collect diverse images with Flickr API for threat detection.
• Preprocess images for model enhancement e.g. resizing, normalizing pixel values
• TensorFlow’s Object Detection API is used, indicating the utilization of pre-trained models for efficient threat detection.
• Design CNN architecture in Keras for object detection. Tools: TensorFlow, Keras
• Train CNN model with transfer learning for accuracy e.g. VGG16

EVALUATION PLAN

• Assess model performance using accuracy metrics.
• Implement cross-validation for validation.
• Use early stopping to prevent overfitting.
• Visualize model predictions for interpretation.