Merge branch 'main' of https://github.com/OpenGVLab/PonderV2 into main

OpenGVLab · Apr 25, 2024 · fa56b05 · fa56b05
2 parents f358d8f + 546b7b9
commit fa56b05
Show file tree

Hide file tree

Showing 5 changed files with 48 additions and 7 deletions.
diff --git a/README.md b/README.md
@@ -10,7 +10,8 @@
 
 [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ponderv2-pave-the-way-for-3d-foundataion/semantic-segmentation-on-scannet)](https://paperswithcode.com/sota/semantic-segmentation-on-scannet?p=ponderv2-pave-the-way-for-3d-foundataion)<br>
 [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ponderv2-pave-the-way-for-3d-foundataion/3d-semantic-segmentation-on-scannet200)](https://paperswithcode.com/sota/3d-semantic-segmentation-on-scannet200?p=ponderv2-pave-the-way-for-3d-foundataion)<br>
-[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ponderv2-pave-the-way-for-3d-foundataion/semantic-segmentation-on-s3dis)](https://paperswithcode.com/sota/semantic-segmentation-on-s3dis?p=ponderv2-pave-the-way-for-3d-foundataion)
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ponderv2-pave-the-way-for-3d-foundataion/semantic-segmentation-on-s3dis)](https://paperswithcode.com/sota/semantic-segmentation-on-s3dis?p=ponderv2-pave-the-way-for-3d-foundataion)<br>
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ponderv2-pave-the-way-for-3d-foundataion/3d-semantic-segmentation-on-scannet-1)](https://paperswithcode.com/sota/3d-semantic-segmentation-on-scannet-1?p=ponderv2-pave-the-way-for-3d-foundataion)
 
 </div>
 
@@ -111,6 +112,10 @@ Finetune PonderV2 on ScanNet semantic segmentation downstream task with PPT:
 # -w: path to checkpoint
 bash scripts/train.sh -g 8 -d scannet -c semseg-ppt-v1m1-0-sc-s3-st-spunet-lovasz-ft -n ponderv2-semseg-ft -w ${PATH/TO/CHECKPOINT}
 ```
+Finetune PonderV2 on ScanNet instance segmentation downstream task using PointGroup:
+```bash
+bash scripts/train.sh -g 4 -d scannet -c insseg-ppt-v1m1-0-pointgroup-spunet-ft -n insseg-pointgroup-v1m1-0-spunet-ft -w ${PATH/TO/CHECKPOINT}
+```
 
 - **Testing**: Test a finetuned model on a downstream task.
 ```bash

diff --git a/configs/s3dis/pretrain-ponder-spunet-v1m1-0-base.py b/configs/s3dis/pretrain-ponder-spunet-v1m1-0-base.py
@@ -185,7 +185,13 @@
         render_semantic=True,
         num_cameras=num_cameras,
         transform=[
-            dict(type="CenterShift", apply_z=True),
+            dict(
+                type="CenterShift",
+                apply_z=True,
+                keys=[
+                    "extrinsic",
+                ],
+            ),
             dict(
                 type="RandomDropout",
                 dropout_ratio=0.8,

diff --git a/configs/scannet/pretrain-ponder-ppt-v1m1-0-sc-s3-st-spunet.py b/configs/scannet/pretrain-ponder-ppt-v1m1-0-sc-s3-st-spunet.py
@@ -267,7 +267,13 @@
                 render_semantic=True,
                 num_cameras=num_cameras,
                 transform=[
-                    dict(type="CenterShift", apply_z=True),
+                    dict(
+                        type="CenterShift",
+                        apply_z=True,
+                        keys=[
+                            "extrinsic",
+                        ],
+                    ),
                     dict(
                         type="RandomDropout",
                         dropout_ratio=0.8,
@@ -366,7 +372,13 @@
                 render_semantic=True,
                 num_cameras=num_cameras,
                 transform=[
-                    dict(type="CenterShift", apply_z=True),
+                    dict(
+                        type="CenterShift",
+                        apply_z=True,
+                        keys=[
+                            "extrinsic",
+                        ],
+                    ),
                     dict(
                         type="RandomDropout",
                         dropout_ratio=0.8,
@@ -464,7 +476,13 @@
                 render_semantic=True,
                 num_cameras=num_cameras,
                 transform=[
-                    dict(type="CenterShift", apply_z=True),
+                    dict(
+                        type="CenterShift",
+                        apply_z=True,
+                        keys=[
+                            "extrinsic",
+                        ],
+                    ),
                     dict(
                         type="RandomDropout",
                         dropout_ratio=0.8,

diff --git a/configs/scannet/pretrain-ponder-spunet-v1m1-0-base.py b/configs/scannet/pretrain-ponder-spunet-v1m1-0-base.py
@@ -203,7 +203,13 @@
         render_semantic=True,
         num_cameras=num_cameras,
         transform=[
-            dict(type="CenterShift", apply_z=True),
+            dict(
+                type="CenterShift",
+                apply_z=True,
+                keys=[
+                    "extrinsic",
+                ],
+            ),
             dict(
                 type="RandomDropout",
                 dropout_ratio=0.8,

diff --git a/configs/structured3d/pretrain-ponder-spunet-v1m1-0-base.py b/configs/structured3d/pretrain-ponder-spunet-v1m1-0-base.py
@@ -209,7 +209,13 @@
         render_semantic=True,
         num_cameras=num_cameras,
         transform=[
-            dict(type="CenterShift", apply_z=True),
+            dict(
+                type="CenterShift",
+                apply_z=True,
+                keys=[
+                    "extrinsic",
+                ],
+            ),
             dict(
                 type="RandomDropout",
                 dropout_ratio=0.8,