TLD

This is a C++ implementation of the Tracking-Learning-Detection (TLD) algorithm.

The code is functional right now, but not stable, i.e. little changes of parameters will lead the result turning worse.

Following is the performance evaluation on TLD dataset measured by Precision / Recall / F-measure.

Sequence	Frames	OpenTLD(MatLab)	My implementation(404dd93302)
1.David	761	1.00 / 1.00 / 1.00	0.968379 / 0.968379 / 0.968379
2.Jumping	313	1.00 / 1.00 / 1.00	1.000000 / 0.990385 / 0.995169
3.Pedestrian 1	140	1.00 / 1.00 / 1.00	1.000000 / 0.992806 / 0.996390
4.Pedestrian 2	338	0.89 / 0.92 / 0.91	0.842857 / 0.445283 / 0.582716
5.Pedestrian 3	184	0.99 / 1.00 / 0.99	0.987097 / 1.000000 / 0.993506
6.Car	945	0.92 / 0.97 / 0.94	0.928339 / 0.996503 / 0.961214
7.Motocross	2665	0.89 / 0.77 / 0.83	0.821918 / 0.849858 / 0.835655
8.Volkswagen	8576	0.80 / 0.96 / 0.87	0.834528 / 0.997275 / 0.908672
9.Carchase	9928	0.86 / 0.70 / 0.77	0.798744 / 0.646189 / 0.714413
10.Panda	3000	0.58 / 0.63 / 0.60	0.630420 / 0.692561 / 0.660031