Added CSV logging for drowsy detection with duration tracking

[ravicoder01]

This PR enhances the Driver Drowsiness Detection System by adding automatic CSV logging whenever drowsiness is detected.

🔥 What’s New
✅ Automatic CSV file creation (blink_log.csv) – Stores important data whenever the system detects the driver is drowsy.
✅ Logs these details:

📅 Date – When drowsiness was detected

⏰ Time – Exact timestamp

🛑 Status – Shows “Drowsy”

⏳ Duration (sec) – How long the driver’s eyes remained closed

✅ Data is only logged when drowsiness is detected (avoids unnecessary entries).

💡 Why this helps
🚗 Safety Tracking: Helps analyze how often a driver becomes drowsy.

      📊 Post-drive Analysis: Data can be reviewed later for improving driver awareness and safety measures.

      📥 Easy Reporting: Since the log is stored in a CSV, it can be directly opened in Excel or Google Sheets for reporting and analysis.

📌 (This log file can be shared with fleet managers or safety officers to monitor driver fatigue patterns.)

[ravicoder01]

Hii @Gagandeep-2003 PR is ready to be reviewed let me know if any changes are required.

[Gagandeep-2003]

@ravicoder01 This blinkDetect.py file is having a conflict. Kindly resolve it. Also mention which issue it is resolving.

[ravicoder01]

@Gagandeep-2003 conflicts has been resolved please review

[Gagandeep-2003]

---

Problems Identified

1. Model Incompatibility

   • The facial landmark model was changed from 70 points to 68 points (shape_predictor_70_face_landmarks.dat → shape_predictor_68_face_landmarks.dat).
   • This change will break compatibility with existing code and pre-trained models.

2. Hardcoded Paths

   • Python executable paths are hardcoded as venv\Scripts\python.exe, which may not work on all systems or OS environments.

3. Unnecessary Code Rewrites

   • Large sections of working code (131 lines) were rewritten, rather than extending the current logic.
   • This makes the PR hard to review and introduces unnecessary merge risk.

4. Large Binary Files in PR

   • Model and video files (.dat, .avi) totaling over 100MB have been added to the repository.

     • This is not recommended. Binary/model files should not be tracked in source control.
     • Best practice is to use GitHub Releases or external storage for these assets.

---

Suggestions to Fix

1. Model

   • Revert to the original 70-landmark model to ensure backward compatibility with the existing codebase.

2. Path Handling

   • Use sys.executable to dynamically reference the current Python interpreter, instead of hardcoded paths.

3. Code Structure

   • Add CSV logging as an additional feature, avoiding major rewrites to the core detection logic.

4. Binary Files

   • Remove all large binary/model/video files from the PR.
   • Instead, provide download instructions or maintain a models/README.md with model links.

5. Documentation

   • Update the documentation or README.md to mention the new CSV logging feature and include usage instructions.

---

[ravicoder01]

Hey @Gagandeep-2003 , I noticed code expects shape_predictor_70_face_landmarks.dat, but README lists the 68-point model. I’ve searched but couldn’t find the 70‑point version. Could you please share it or allow me to update code for 68‑point model?