- Added instructions for running the FastAPI server (uvicorn)
- Explained various API endpoints such as point cloud generation and height estimation
- Included usage details for the orchestrator system
- Described the demo.py script integrating YOLOv8 for real-time person detection and height estimation
- Updated calibration instructions for rectifying images and generating point clouds

feat(orchestrator + threading): integrate YOLO detection with orchestrator and implement threading for smooth processing

- Integrated YOLO person detection with the orchestrator for real-time height estimation.
- Implemented threading to handle the height estimation in parallel with the video feed, preventing pauses during processing.
- Adjusted the orchestrator to work seamlessly with live camera input for height estimation.
- Added a time interval to control when the height estimation is processed, avoiding constant processing.
- The video feed remains uninterrupted while the orchestrator processes the required computations in a separate thread.

- Implemented Orchestrator class to manage module execution based on the given requirement (dense, nodense, features, height).
- Added capability to dynamically set new images and requirements.
- Integrated profile loading and image rectification for point cloud generation modules.
- Created demo_orchestrator.py to handle user input and execute corresponding modules in a loop.

feat(detection): integrated YOLOv8 person detection with webcam support

- Added detect_person_webcam function using YOLOv8 for real-time person detection.
- Implemented a live webcam detection loop displaying bounding boxes around detected people.
- Set confidence threshold to 0.8 for higher accuracy in detections.

- Implemented the generate_filtered_point_cloud_with_features function to generate point clouds with keypoints and feature extraction based on YOLOv8.
- Added the corresponding API endpoint to allow users to upload stereo images and receive filtered point clouds, keypoints, and feature data.
- Enhanced the functionality to visualize point clouds and keypoints, and return the extracted features for further processing.
- Included support for handling normalization and disparity methods in the generation process.

- Implemented a new GET endpoint `/get_profile/{profile_name}`.
- Returns the JSON representation of a profile based on the provided `profile_name`.
- Raises a 404 HTTP exception if the profile is not found.