CharlesGaydon
commited on
Commit
•
10b8275
1
Parent(s):
e4594ae
Update README.md
Browse files
README.md
CHANGED
@@ -45,7 +45,7 @@ model-index:
|
|
45 |
<h1>FRACTAL-LidarHD_7cl_randlanet</h1>
|
46 |
<p>The general characteristics of this specific model <strong>FRACTAL-LidarHD_7cl_randlanet</strong> are :</p>
|
47 |
<ul style="list-style-type:disc;">
|
48 |
-
<li>Trained with the FRACTAL dataset</li>
|
49 |
<li>Aerial lidar point clouds, colorized with rgb + near-infrared, with high point density (~40 pts/m²)</li>
|
50 |
<li>RandLa-Net architecture as implemented in the Myria3D library</li>
|
51 |
<li>7 class nomenclature : [other, ground, vegetation, building, water, bridge, permanent structure]</li>
|
@@ -157,12 +157,39 @@ The model was obtained for num_epoch=21 with corresponding val_loss=0.112.
|
|
157 |
|
158 |
#### Testing Data
|
159 |
|
|
|
|
|
|
|
|
|
160 |
#### Metrics
|
161 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
162 |
|
163 |
### Results
|
164 |
|
165 |
-
Samples of results
|
|
|
|
|
|
|
|
|
|
|
|
|
166 |
|
167 |
|
168 |
---
|
|
|
45 |
<h1>FRACTAL-LidarHD_7cl_randlanet</h1>
|
46 |
<p>The general characteristics of this specific model <strong>FRACTAL-LidarHD_7cl_randlanet</strong> are :</p>
|
47 |
<ul style="list-style-type:disc;">
|
48 |
+
<li>Trained with the FRACTAL dataset for the semantic segmentation of Lidar HD point clouds</li>
|
49 |
<li>Aerial lidar point clouds, colorized with rgb + near-infrared, with high point density (~40 pts/m²)</li>
|
50 |
<li>RandLa-Net architecture as implemented in the Myria3D library</li>
|
51 |
<li>7 class nomenclature : [other, ground, vegetation, building, water, bridge, permanent structure]</li>
|
|
|
157 |
|
158 |
#### Testing Data
|
159 |
|
160 |
+
The model was evaluated on the 10,000 data patches of the test set of the FRACTAL dataset,
|
161 |
+
that are independant from train and val patches, and sampled from distinct areas in the five spatial domains of the dataset.
|
162 |
+
The diversity of landscapes and scenes of the test set should closely match the one of the train and val sets.
|
163 |
+
|
164 |
#### Metrics
|
165 |
|
166 |
+
The **FRACTAL-LidarHD_7cl_randlanet** model achieves a performance of mIoU=77.2% and OA=XXX%.
|
167 |
+
|
168 |
+
The following table fives the class-wise metrics:
|
169 |
+
|
170 |
+
TODO: add IoU, and other metrics if available.
|
171 |
+
|
172 |
+
|
173 |
+
The following illustration gives the resulting confusion matrix :
|
174 |
+
* Left : normalised acording to rows: rows sum at 100% and the **recall** is on the diagonal of the matrix
|
175 |
+
* Right : normalised acording to columns: columns sum at 100% and the **precision** is on the diagonal of the matrix
|
176 |
+
|
177 |
+
|
178 |
+
<div style="position: relative; text-align: center;">
|
179 |
+
<p style="margin: 0;">Normalized Confusion Matrices. (a) Recall, (b) Precision)</p>
|
180 |
+
<img src="FRACTAL-LidarHD_7cl_randlanet-recall_confusion_matrix.excalidraw.png" alt="Confusion matrices" style="width: 70%; display: block; margin: 0 auto;"/>
|
181 |
+
</div>
|
182 |
+
|
183 |
|
184 |
### Results
|
185 |
|
186 |
+
Samples of results.
|
187 |
+
From test patches with at least 10k points (i.e. at least 4 pts/m²), we sample without cherry-picking, with the following criteria:
|
188 |
+
* (1) WATER & BRIDGE
|
189 |
+
* (2) BUILD_GREENHOUSE
|
190 |
+
* (3) OTHER_PARKING
|
191 |
+
* (4) HIGHSLOPE1
|
192 |
+
* (5) URBAN
|
193 |
|
194 |
|
195 |
---
|