Advanced Features and Tracking

Buy in Store

This chapter describes the system's automatic tracking features in detail, including color tracking, face tracking, and QR code tracking, as well as the target lock mechanism and parameter tuning.

---

Automatic Tracking Modes Overview

The system supports three automatic tracking modes:

1. Color tracking: track objects of a specified color
2. Face tracking: track faces
3. QR code tracking: track QR codes

---

Color Tracking

Color Selection

The system supports tracking of several colors:

• Red
• Green
• Blue You can switch colors with keys in the program:
• X: select red
• Y: select green
• Z: select blue

Color Tracking Workflow

1. Press C to connect the gimbal
2. Press 2 to enter color tracking mode
3. Place the object of the target color in the center of the frame
4. Press T to lock the target
5. Move the target and observe the gimbal following it

Simple Color Tracking Example

You can also use the simple color tracking example program:

python examples/04_color_track_simple.py --camera 0 --port COM3 --color red

This program provides the most basic color tracking functionality and is good for learning.

---

Face Tracking

How Face Tracking Works

The system uses OpenCV's Haar cascade classifier for face detection. Once a face is detected, the system automatically calculates the target position and controls the gimbal to follow it.

Face Tracking Workflow

1. Press C to connect the gimbal
2. Press 1 to enter face tracking mode
3. Place your face in the center of the frame
4. Press T to lock the target
5. Move your face; the gimbal follows automatically

Tips for Improving Face Detection

• Keep the lighting adequate and avoid backlight
• Face the camera directly
• Keep an appropriate distance (1-3 m recommended)
• Avoid scenes with multiple faces, or use the lock mechanism to keep the tracking target fixed

---

QR Code Tracking

QR code tracking mode uses OpenCV's QRCodeDetector for QR code recognition and localization. It works in a similar way to the previous two tracking modes:

1. Connect the gimbal and enter QR code tracking mode
2. Place the QR code in the center of the frame and press T to lock
3. Move the QR code and observe the gimbal following it

---

Target Lock Mechanism

What Locking Does

The target lock mechanism is a key feature of the system. It:

• Records the target's center position and size at the moment of locking
• Prefers the target closest to the lock point when multiple targets appear
• Prevents the target from jumping around frequently, keeping tracking stable

Locking Workflow

1. Place the target object in the center of the frame
2. Press T to lock
3. After a successful lock, the system prioritizes the target most similar to the one locked
4. Press S to cancel the lock and stop tracking

Target Selection Logic After Locking

When selecting a target after locking, the system considers two factors:

• Distance: how close the target center is to the lock point (weight 70%)
• Size: how similar the target's size is to the locked size (weight 30%)
• The system tracks the target with the highest combined score

---

Tuning Tracking Control Parameters

The following tunable parameters are in src/trackers/tracking_controller.py:

Parameter	Default	Description
kp_pan	0.08	Proportional gain for pan tracking
kp_tilt	0.12	Proportional gain for tilt tracking
dead_zone	30	Dead zone (pixels); the gimbal does not move within this range
min_move_interval	0.15	Minimum move interval (seconds); limits how often the gimbal moves

How to Adjust the Parameters

• Tracking too slow: increase kp_pan and kp_tilt
• Tracking too sensitive, causing jitter: decrease kp_pan and kp_tilt, increase min_move_interval, or increase dead_zone
• Frequent small adjustments causing jitter: increase dead_zone
• Direction reversed: change the sign of delta_pan or delta_tilt in the calculate_move method

---

Complete Tracking Example

Here is a complete usage example:

1. Start the program:

python examples/auto_tracking_demo.py --camera 0 --port COM3 --color red

1. Press C to connect the gimbal
2. Press 2 to select color tracking mode
3. Place the red object in the center of the frame
4. Press T to lock the target
5. Move the object and observe the gimbal following it
6. To switch to green, press Y and then press T again to lock
7. Press S to stop tracking and press R to return to center
8. Press Q to quit

---

Advanced Development Tips

If you need custom features or further development, refer to:

• src/sc_servo.py: low-level servo communication
• src/gimbal.py: gimbal control
• src/trackers/tracking_controller.py: tracking controller
• src/detectors/: various target detectors