Table 1.

F1 integral score (mean ± standard deviation) results for three methods across six datasets. Statistical significance was determined by one-sided Wilcoxon signed-rank test (deepBlink greater than) with **P < 1e⁻², ****P < 1e⁻⁴. Size of test set: 129 smFISH, 105 SunTag, 64 Particle, 240 Microtubule, 240 Receptor and 240 Vesicle. Bold signifies the highest score

		TrackMate		DetNet		deepBlink
Real	smFISH	0.865 ± 0.177	****	0.442 ± 0.250	****	0.905 ± 0.145
	SunTag	0.652 ± 0.328	**	0.036 ± 0.040	****	0.712 ± 0.279
Synthetic	Particle	0.941 ± 0.008	****	0.828 ± 0.014	****	0.944 ± 0.008
	Microtubule	0.380 ± 0.228	****	0.298 ± 0.149	****	0.637 ± 0.261
	Receptor	0.606 ± 0.344	****	0.517 ± 0.258	****	0.682 ± 0.310
	Vesicle	0.459 ± 0.198	****	0.520 ± 0.245	****	0.732 ± 0.259
	Average	0.651 ± 0.214		0.440 ± 0.159		0.769 ± 0.210