Downstream Semantic Segmentation Model for Low-Level Surface Crack Detection

Table 3

Comparing our downstream models built upon DeepLabV3-ResNet101 with other works of deep-learning-based low-level crack detection on the same test datasets; the flexibility value of  = 2 is used in all evaluations. While ODS F-measure is reported with its corresponding precision (P) and recall (R) as written underneath, OIS F-measure is reported together with its standard deviation (written with a leading ± symbol) showing the dispersion of the optimal image-scale F-measure in a particular test dataset. For each training strategy of us, the best (maximum) value among the four loss functions is highlighted in bold.

	Modified U-Net [6]	DeepCrack [28]	Our downstream models built upon DeepLabV3-ResNet101
			Trained from pretrained weights				Trained from random weights
			CrackTree260
OIS F-measure	0.3981	–	0.3081	0.8520	0.8596	0.8752	0.3031	0.8542	0.8577	0.8705
			.0919	.0468	.0521	.0312	.0594	.0227	.0241	.0230
ODS F-measure	0.3940	–	0.2218	0.8354	0.8408	0.8553	0.2721	0.8440	0.8449	0.8530
			@P .1325	@P .7706	@P .7835	@P .8049	@P .1789	@P .7777	@P .7870	@P .8056
			@R .6793	@R .9120	@R .9073	@R .9124	@R .5684	@R .9226	@R .9120	@R .9064
AP	0.1683	–	0.1317	0.8347	0.8387	0.8879	0.1764	0.8595	0.8820	0.8913
CRKWH100
OIS F-measure	0.3350	0.9170	0.1117	0.5321	0.6107	0.5447	0.1681	0.8131	0.8020	0.8111
			.1208	.3660	.3404	.3595	.0983	.1282	.1315	.1276
ODS F-measure	0.3483	0.9095	0.1407	0.5226	0.4480	0.6612	0.0951	0.8166	0.8029	0.8011
			@P 0.0848	@P 0.3934	@P 0.3072	@P 0.5338	@P 0.0595	@P 0.7530	@P 0.7416	@P 0.7429
			@R 0.4128	@R 0.7785	@R 0.8277	@R 0.8686	@R 0.2369	@R 0.8919	@R 0.8752	@R 0.8693
AP	0.1242	0.9315	0.0642	0.3490	0.2629	0.3938	0.0421	0.8362	0.8215	0.8239
CrackLS315
OIS F-measure	0.2517	0.8671	0.1371	0.6844	0.7191	0.6761	0.1832	0.7315	0.7177	0.7217
			0.1009	0.2237	0.1924	0.2467	0.1095	0.1734	0.1808	0.1832
ODS F-measure	0.2648	0.8449	0.0791	0.7143	0.7332	0.6698	0.1143	0.7446	0.7255	0.7325
			@P 0.0560	@P 0.6322	@P 0.6587	@P 0.5656	@P 0.0628	@P 0.6722	@P 0.6597	@P 0.6663
			@R 0.1345	@R 0.8208	@R 0.8268	@R 0.8210	@R 0.6413	@R 0.8345	@R 0.8059	@R 0.8133
AP	0.1004	0.8772	0.0351	0.6378	0.6893	0.5178	0.0512	0.7205	0.7096	0.7138
Stone331
OIS F-measure	0.4090	0.8751	0.0605	0.0480	0.0532	0.0564	0.0453	0.6683	0.7587	0.6578
			0.0687	0.0200	0.0273	0.0406	0.0309	0.1719	0.1348	0.1976
ODS F-measure	0.4333	0.8559	0.0313	0.0340	0.0358	0.0288	0.0277	0.6829	0.7572	0.6822
			@P 0.0159	@P 0.0173	@P 0.0182	@P 0.0146	@P 0.0140	@P 0.6036	@P 0.7075	@P 0.6264
			@R 0.8917	@R 0.9817	@R 0.9662	@R 0.9784	@R 1.0000	@R 0.7862	@R 0.8144	@R 0.7488
AP	0.2374	0.8883	0.0126	0.0123	0.0130	0.0125	0.0098	0.6686	0.7697	0.6653