OpenDroneMap / ODMSemantic3D

An open photogrammetry dataset of classified 3D point clouds for automated semantic segmentation. CC BY-SA 4.0
Other
15 stars 7 forks source link

Added sheffield park point cloud #36

Closed HeDo88TH closed 1 year ago

HeDo88TH commented 1 year ago

As title says!

u4gbot commented 1 year ago
sheffield_park_small Overall accuracy from 23.63% to 87.62% | building | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 51.70% | 70.70% | 65.80% | 0.68 | | After | 54.78% | 55.38% | 98.06% | 0.71 | | Diff | 3.08% | -15.32% | 32.26% | 0.03 | | human_made_object | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 1.43% | 1.50% | 21.78% | 0.03 | | After | 0.00% | 0.00% | 0.00% | N/A | | Diff | -1.43% | -1.50% | -21.78% | N/A | | low_vegetation | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 18.60% | 99.98% | 18.60% | 0.31 | | After | 58.84% | 99.67% | 58.96% | 0.74 | | Diff | 40.24% | -0.31% | 40.36% | 0.43 | | ground | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 0.44% | 0.44% | 86.19% | 0.01 | | After | 89.51% | 90.00% | 99.39% | 0.94 | | Diff | 89.06% | 89.56% | 13.20% | 0.94 |
brighton_beach_small Overall accuracy from 95.76% to 82.38% | building | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 22.14% | 53.03% | 27.53% | 0.36 | | After | 25.25% | 98.97% | 25.32% | 0.40 | | Diff | 3.11% | 45.94% | -2.22% | 0.04 | | human_made_object | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 36.89% | 37.05% | 98.85% | 0.54 | | After | 0.00% | 0.00% | N/A | N/A | | Diff | -36.89% | -37.05% | N/A | N/A | | low_vegetation | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 23.78% | 29.40% | 55.45% | 0.38 | | After | 16.38% | 98.94% | 16.41% | 0.28 | | Diff | -7.40% | 69.54% | -39.04% | -0.10 | | ground | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 96.07% | 98.65% | 97.36% | 0.98 | | After | 82.32% | 82.32% | 100.00% | 0.90 | | Diff | -13.75% | -16.32% | 2.64% | -0.08 |

Average (2 datasets)

Overall accuracy from 59.70% to 85.00%

building Accuracy Recall Precision F1
Before 36.92% 61.87% 46.67% 0.52
After 40.02% 77.18% 61.69% 0.56
Diff 3.10% 15.31% 15.02% 0.03
human_made_object Accuracy Recall Precision F1
Before 19.16% 19.28% 60.32% 0.28
After 0.00% 0.00% 0.00% 0.00
Diff -19.16% -19.28% -60.32% -0.28
low_vegetation Accuracy Recall Precision F1
Before 21.19% 64.69% 37.02% 0.35
After 37.61% 99.30% 37.68% 0.51
Diff 16.42% 34.62% 0.66% 0.16
ground Accuracy Recall Precision F1
Before 48.26% 49.54% 91.77% 0.49
After 85.91% 86.16% 99.69% 0.92
Diff 37.66% 36.62% 7.92% 0.43