0
# Tracking
1

2
Multi-object tracking algorithms for maintaining object identities across video frames.
3

4
## Capabilities
5

6
### ByteTrack Algorithm
7

8
```python { .api }
9
class ByteTrack:
10
    """
11
    Multi-object tracking using the ByteTracker algorithm.
12
    
13
    Args:
14
        track_thresh (float): Detection confidence threshold for tracking
15
        track_buffer (int): Number of frames to keep lost tracks
16
        match_thresh (float): Matching threshold for track association
17
        frame_rate (int): Video frame rate for track management
18
    """
19
    def __init__(
20
        self,
21
        track_thresh: float = 0.25,
22
        track_buffer: int = 30,
23
        match_thresh: float = 0.8,
24
        frame_rate: int = 30,
25
    ): ...
26

27
    def update_with_detections(self, detections: Detections) -> Detections:
28
        """
29
        Update tracker with new detections and return tracked results.
30
        
31
        Args:
32
            detections (Detections): Current frame detections
33
            
34
        Returns:
35
            Detections: Detections with assigned tracker_id values
36
        """
37
```
38

39
## Usage Example
40

41
```python
42
import supervision as sv
43
from ultralytics import YOLO
44

45
# Setup model and tracker
46
model = YOLO("yolov8n.pt") 
47
tracker = sv.ByteTrack()
48

49
# Process video frames
50
for frame in sv.get_video_frames_generator("video.mp4"):
51
    # Get detections
52
    results = model(frame)[0]
53
    detections = sv.Detections.from_ultralytics(results)
54
    
55
    # Update tracker
56
    detections = tracker.update_with_detections(detections)
57
    
58
    # Now detections have tracker_id assigned
59
    print("Tracked objects:", len(detections))
60
```