Fencer Trackinng
X
1
Position
After detecting the fencers, the model must continuously track their movements. This starts by determining the coordinates of the fencers as they appear on screen. To achieve this, an "anchor point" is designated between the feet of each fencer using the YOLOv8 model. The coordinates of this anchor point are recorded as the variables X and Y, while the relative "size" of each fencer is captured using two additional variables, shape_x and shape_y.
2
Algorithm
Between frames, fencers' positions shift slightly. To track them accurately, the model compares their coordinates in the current frame with those from the previous one. The final algorithm establishes a verification range using size data:
(x ± 0.25 * shape_x, y ± 0.25 * shape_y).
3
Tracking
If a fencer's position falls within this range relative to the previous frame's coordinates, they are identified as the same individual and tracked accordingly. This process is repeated for both detected fencers in every frame, ensuring reliable tracking even with rapid or drastic movements.
