[Prayas340]

Description

This Pull Request addresses critical computer vision bugs, filtering defects, missing temporal stabilization, and CLI limitations in the Finding_Lanes project:

1. Fix Multi-Channel Color Corruption in roi() Mask:

   • Resolved bug where cv2.fillPoly(mask, polygons, 255) applied integer scalar 255, which only populated channel 0 (blue) and wiped out channels 1 (green) and 2 (red) on 3-channel color images.
   • Dynamically adapts mask fill value: (255,) * channels for multi-channel frames, preserving full RGB/BGR color data.

2. Fix Unconstrained Slope Averaging & Cross-Lane Contamination:

   • In average_lines_intercept(), segments were previously segregated solely based on slope signs (slope < -min_slope vs slope > min_slope).
   • Added spatial midpoint constraints (line_mid_x < midpoint for left lanes, line_mid_x > midpoint for right lanes). This prevents negative-slope noise markings, shadows, or guardrails on the right half of the road from contaminating the left lane model.

3. Add Temporal Lane Tracking & Smoothing (LaneTracker):

   • Introduced the LaneTracker class using an Exponential Moving Average (EMA) filter to eliminate frame-to-frame video jitter.
   • Built in a dropout retention buffer (max_missed_frames=15) so that brief gaps between dashed road lane markings no longer cause the lane line to flicker or disappear.

4. Add Drivable Lane Corridor Visualization (--fill-lane):

   • Added --fill-lane CLI flag and parameter to render a semi-transparent filled polygon between the left and right lane boundaries, highlighting the vehicle's drivable lane corridor.

5. Fix Gamma Offset (+1 Brightness Shift) in cv2.addWeighted:

   • Changed gamma=1.0 to standard gamma=0.0 in process_frame() to avoid artificially washing out and brightening image frames.

6. Fix CLI Exit Codes and Error Handling:

   • Replaced silent 0-exit returns with sys.exit(1) when input files or video streams are invalid or not found, ensuring compatibility with CI/CD and automated scripting pipelines.
   • Synchronized GUI video playback delay with native video FPS.

7. Add Dependencies Specification (requirements.txt):

   • Created requirements.txt specifying opencv-python>=4.5.0, numpy>=1.20.0, and matplotlib>=3.3.0.

8. Comprehensive Automated Unit Test Suite (test_lanes.py):

   • Expanded unit tests from 14 to 22 test cases, testing 3-channel color preservation, spatial cross-contamination rejection, EMA smoothing, dropout retention, drivable corridor rendering, and missing-file error handling.

9. Documentation Overhaul (README.md):

   • Updated documentation with new CLI options (--fill-lane, --no-smooth, --no-show), installation instructions, and updated project structure.

---

Fixes #603

• Issue_no: [Finding_Lanes] Fix 3-channel ROI color zeroing, cross-lane contamination, add LaneTracker smoothing & drivable corridor #603
• Not applicable

Type of change

Choosing one or more options from the following as per the nature of your Pull request:

• New project insert
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Documentation Update

Project

• Your Project Name: Finding Lanes - Computer Vision Lane Detection

• Short Description: Real-time road lane detection and drivable corridor visualization using OpenCV and NumPy, featuring Canny edge detection, dynamic ROI, Hough transform, slope-intercept regression, and temporal EMA smoothing.

Checklist:

Please tick all the boxes that are fulfilled by your Pull Request.

• I have named my files and folder, according to this project's guidelines.
• My code follows the style guidelines of this project.
• I have commented on my code, particularly in hard-to-understand areas.
• I have created a helpful and easy to understand README.md, according to the given README_TEMPLATE.
• My changes do not produce any warnings.