Skip to content

Commit 8b1e9e5

Browse files
angeladaiangeladai
committed
Merge branch 'master' of https://github.com/ScanNet/ScanNet 
2 parents e2bf12b + abad773 commit 8b1e9e5

1 file changed

Lines changed: 5 additions & 3 deletions

File tree

‎BenchmarkScripts/README.md‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -5,9 +5,9 @@ We provide scripts for evaluation of the tasks in the [ScanNet benchmark](http:/
55
These scripts have been developed and tested with Python 2.7.
66

77
## Evaluation Scripts
8-
The evaluation for both 2D and 3D semantic label and instance segmentation tasks are adapted from the [CityScapes evaluation](https://github.com/mcordts/cityscapesScripts), and provided in [2d_evaluation](2d_evaluation) and [3d_evaluation](3d_evaluation), respectively.
8+
The evaluation for both 2D and 3D semantic label and instance segmentation tasks are adapted from the [CityScapes evaluation](https://github.com/mcordts/cityscapesScripts), and provided in [2d_evaluation](2d_evaluation) and [3d_evaluation](3d_evaluation), respectively. The evaluation for the scene type classification task is provided in [scene_type_evaluation](scene_type_evaluation).
99

10-
Note that visualization and data preparation scripts depend on the ScanNet label mapping file `scannetv2-labels.combined.tsv` to convert labels into the NYUv2 40-label set, which we use for evaluation.
10+
Note that visualization and data preparation scripts for semantic instance and label prediction depend on the ScanNet label mapping file `scannetv2-labels.combined.tsv` to convert labels into the NYUv2 40-label set, which we use for evaluation.
1111

1212
Please see the [benchmark evaluation documentation](http://kaldir.vc.in.tum.de/scannet_benchmark/documentation) for more information about evaluation data formats and label sets.
1313

@@ -19,4 +19,6 @@ Please see the [benchmark evaluation documentation](http://kaldir.vc.in.tum.de/s
1919
* 3D:
2020
* [3d_helpers](3d_helpers) contains helper scripts for exporting a train scan into the 3D evaluation format for semantic instance and label segmentation.
2121
* `3d_helpers/visualize_labels_on_mesh` can be used to visualize semantic labels on a mesh (assumes NYUv2 40 labels).
22-
22+
23+
* Scene Types:
24+
* `scene_type_helpers/get_scene_type_for_scan` can be used to get the scene type id from the info `<scanId>.txt` file for a scan in the ScanNet release.

0 commit comments

Comments
 (0)