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 jwcate point cloud #38

Closed HeDo88TH closed 1 year ago

HeDo88TH commented 1 year ago

As title says!

u4gbot commented 1 year ago
brighton_beach_small Overall accuracy from 98.01% to 92.26% | building | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 52.60% | 88.05% | 56.65% | 0.69 | | After | 10.34% | 99.56% | 10.34% | 0.19 | | Diff | -42.27% | 11.51% | -46.31% | -0.50 | | ground | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 98.31% | 99.39% | 98.91% | 0.99 | | After | 92.45% | 93.34% | 98.98% | 0.96 | | Diff | -5.87% | -6.05% | 0.07% | -0.03 | | low_vegetation | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 63.60% | 70.04% | 87.36% | 0.78 | | After | 60.63% | 66.94% | 86.53% | 0.75 | | Diff | -2.97% | -3.10% | -0.82% | -0.02 | | human_made_object | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 34.65% | 43.46% | 63.08% | 0.51 | | After | 26.20% | 50.70% | 35.16% | 0.42 | | Diff | -8.44% | 7.24% | -27.91% | -0.10 |
sheffield_park_small Overall accuracy from 89.67% to 91.66% | building | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 64.63% | 65.61% | 97.74% | 0.79 | | After | 81.49% | 83.50% | 97.12% | 0.90 | | Diff | 16.85% | 17.89% | -0.62% | 0.11 | | ground | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 91.08% | 91.61% | 99.36% | 0.95 | | After | 91.12% | 92.07% | 98.89% | 0.95 | | Diff | 0.04% | 0.45% | -0.47% | 0.00 | | low_vegetation | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 63.53% | 99.14% | 63.88% | 0.78 | | After | 68.79% | 98.92% | 69.32% | 0.82 | | Diff | 5.26% | -0.22% | 5.43% | 0.04 | | human_made_object | Accuracy | Recall | Precision | F1 | | -------------------- | -------- | -------- | --------- | -------- | | Before | 0.00% | 0.00% | 0.00% | N/A | | After | 0.65% | 0.70% | 8.11% | 0.01 | | Diff | N/A | N/A | N/A | N/A |

Average (2 datasets)

Overall accuracy from 93.84% to 91.96%

building Accuracy Recall Precision F1
Before 58.62% 76.83% 77.20% 0.74
After 45.91% 91.53% 53.73% 0.54
Diff -12.71% 14.70% -23.46% -0.19
ground Accuracy Recall Precision F1
Before 94.70% 95.50% 99.14% 0.97
After 91.79% 92.70% 98.94% 0.96
Diff -2.91% -2.80% -0.20% -0.02
low_vegetation Accuracy Recall Precision F1
Before 63.56% 84.59% 75.62% 0.78
After 64.71% 82.93% 77.92% 0.78
Diff 1.15% -1.66% 2.30% 0.01
human_made_object Accuracy Recall Precision F1
Before 17.32% 21.73% 31.54% 0.26
After 13.43% 25.70% 21.64% 0.21
Diff -3.90% 3.97% -9.90% -0.04