Women Safety Analytics

---

Women Safety Analytics is a data-driven project aimed at analyzing and visualizing key factors affecting women's safety. The project leverages machine learning and data analytics techniques to extract meaningful insights from crime and safety-related datasets, enabling better understanding and awareness of women's safety issues.

Key Features:

• Data Analysis: In-depth exploration of datasets related to women's safety.
• Visualization: Intuitive and impactful visualizations for better understanding of patterns and trends.
• Machine Learning Models: Predictive analysis to identify high-risk areas or situations.
• Interactive Dashboards: Engaging tools to help users interact with the data insights.

Goals:

• To identify trends in women's safety-related data.
• To assist policymakers, law enforcement, and NGOs in understanding and addressing safety concerns.

Models Used:

1. Gender Classification Model: This model distinguishes between male and female individuals in real-time video feeds, enabling the system to monitor and analyze gender distribution in public areas.

2. Violence Detection Model: Trained to recognize violent behaviors or actions, this model analyzes live surveillance footage to detect potential threats, facilitating prompt intervention.

3. Pose Estimation Model: This model assesses human body postures to identify distress signals or unusual movements that may indicate a person in danger, enhancing the system's ability to detect and respond to emergencies.

4. Facial Emotion Recognition Model: By analyzing facial expressions, this model detects emotions such as fear or distress, providing additional context to assess potential safety threats.

Workflow of the Women Safety Analytics System 🚦

1. Data Input

   • Video Feed/Surveillance Data: The system processes real-time video streams from public or private surveillance cameras.
   • Image Data: Still images are used for specific analysis tasks.

2. Preprocessing

   • Frames are extracted from video streams for further processing.
   • Images are resized, normalized, and augmented (if necessary) to improve model performance.
   • Background noise and irrelevant data are filtered out.

3. Model 1: Gender Classification

   • The first step identifies the gender of individuals in the scene.
   • Helps determine if more women are present in specific areas for focused monitoring.

4. Model 2: Pose Estimation

   • Analyzes the posture of individuals in the frame.
   • Detects suspicious or distress-related body movements (e.g., a person falling or showing signs of struggle).

5. Model 3: Facial Emotion Recognition

   • Evaluates facial expressions to detect fear, distress, or other concerning emotions.
   • Assigns risk scores based on detected emotions.

6. Model 4: Violence Detection

   • Scans the environment for violent actions or sudden aggressive movements.
   • Triggers alerts if violence is detected in the monitored area.

7. Integration and Decision Making

   • Outputs from all models are aggregated and analyzed.
   • A safety risk score is calculated for the monitored area.

8. Alerts and Reporting

   • If the risk score crosses a predefined threshold:
     • An alert is sent to authorities or security personnel.
     • The exact location and a snapshot of the event are highlighted.
   • Generates logs and reports for further analysis.
   • Includes visualizations of gender distribution, risk zones, and detected events.

---