A Light-Weight Neural Network for Fast and Interactive Edge Detection in 3D Point Clouds

Statistical analysis

Indicators: median value per method

Click on a column header to change sorting.

Search:

Method	Precision	Recall	MCC	TP	FP	TN	FN	F1	Accuracy	Selectivity	IoU
CNN (trained on Default)	0.468	0.928	0.570	0.080	0.906	0.920	0.094	0.611	0.911	0.915	0.440
PCED-2C (trained on Default)	0.462	0.623	0.455	0.081	0.658	0.919	0.342	0.510	0.868	0.936	0.342
PCED (trained on Default - P = Sharp+Smooth)	0.349	0.898	0.444	0.122	0.962	0.878	0.038	0.489	0.814	0.822	0.323
CNN (trained on Default - P = Sharp+Smooth)	0.299	0.945	0.429	0.142	0.986	0.858	0.014	0.446	0.802	0.787	0.287
FC (trained on Default - P = Sharp+Smooth)	0.313	0.909	0.428	0.119	0.959	0.881	0.041	0.461	0.793	0.778	0.300
PCED-2C (trained on Shrec)	0.441	0.872	0.426	0.112	0.900	0.888	0.100	0.522	0.874	0.873	0.353
FC-2C (trained on Default)	0.391	0.780	0.421	0.101	0.839	0.899	0.161	0.479	0.846	0.872	0.315
PCED-2C (trained on ABC)	0.479	0.621	0.394	0.067	0.704	0.933	0.296	0.561	0.901	0.936	0.390
FC-2C (trained on Shrec)	0.406	0.715	0.392	0.116	0.835	0.884	0.165	0.480	0.849	0.887	0.316
FC (trained on Default)	0.392	0.538	0.391	0.074	0.650	0.926	0.350	0.455	0.853	0.880	0.294
Covariance Analysis	0.434	0.449	0.390	0.061	0.506	0.939	0.494	0.442	0.876	0.922	0.284
FC-2C (trained on ABC)	0.389	0.693	0.369	0.077	0.753	0.923	0.247	0.438	0.846	0.870	0.280
PCED (trained on Default)	0.495	0.377	0.344	0.047	0.290	0.953	0.710	0.437	0.916	0.956	0.280
Edge-aware Point set Consolidation Network (Pre-trained)	0.365	0.503	0.341	0.060	0.639	0.940	0.361	0.397	0.845	0.893	0.248
PCED-2C (trained on EC)	0.254	0.747	0.285	0.124	0.919	0.876	0.081	0.365	0.676	0.648	0.223
CNN-2C (trained on ABC)	0.271	0.496	0.234	0.062	0.662	0.938	0.338	0.315	0.818	0.857	0.187
PCPNet (trained on ABC)	0.399	0.250	0.211	0.030	0.408	0.970	0.592	0.310	0.880	0.943	0.183
CNN-2C (trained on Shrec)	0.207	0.881	0.202	0.176	0.956	0.824	0.044	0.248	0.622	0.608	0.142
Feature Edge Extraction	0.191	0.527	0.151	0.077	0.826	0.923	0.174	0.278	0.727	0.771	0.160
PCPNet (trained on Default)	0.000	0.000	0.000	0.000	0.000	1.000	1.000	0.009	0.893	1.000	0.000

Showing 1 to 20 of 20 entries

Indicators distributions per method

This plot shows, for each method and training set, the distributions of several indicators for the whole dataset. Use the drop-down menue below to change the indicator (both the description and the plot should be updated). Note that the plot is interactive: change zoom, range and access method full name using mouse interaction.
Current indicator:

Precision (also denoted positive predictive value -- PPV) measures the proportion of positive identifications that was actually correct (higher is better). It is defined as:

precision = \frac{T P}{T P + F P}

Bad (indicator 25% smallest values) and good (indicator 10% higher values) areas are highlighted in the plot in red and green, respectively. Efficient methods should have few samples in the red area (the lower, the better), while having most of its samples in the green area (the higher, the better).

Precision/Recall

This plot shows a representative score per approach: each method is shown as point sample, such that (select mean or median below):

Its position is the mean precision/recall score for the whole dataset,
Its position is the median precision/recall score for the whole dataset.

The plot is interactive: change zoom, range and access method full name using mouse interaction.

Visualization type:

Precision/Recall distributions - Methods comparisons

These graphs show the precision/recall values as scatter plot (each 3D model is a sample) and its associated density function for the whole dataset.

Left dataset:

Right dataset:

A Light-Weight Neural Network for Fast and Interactive Edge Detection in 3D Point Clouds

Feature Curve Extraction on Triangle Meshes (SHREC'19) - Sharp edges

Examples of 3d models

Model description

Legend

Interactive WebGL viewer

Statistical analysis

Indicators: median value per method

Indicators distributions per method

Precision/Recall

Precision/Recall distributions - Methods comparisons