Update testing scripts and README.md

README.md

@@ -127,6 +127,13 @@ cd demo
 </details>
 
 ### Reproduce the testing results with pytorch-lightning
+You need to setup the testing subsets of ScanNet and MegaDepth first. We create symlinks from the previously downloaded datasets to `data/{{dataset}}/test`.
+
+```shell
+# set up symlinks
+ln -s /path/to/scannet-1500-testset/* /path/to/LoFTR/data/scannet/test
+ln -s /path/to/megadepth-1500-testset/* /path/to/LoFTR/data/megadepth/test
+```
 
 ```shell
 conda activate loftr
@@ -142,7 +149,7 @@ For visualizing the results, please refer to `notebooks/visualize_dump_results.i
 <br/>
 
 
-### Image pair info for training on ScanNet
+<!-- ### Image pair info for training on ScanNet
 You can download the data at [here](https://drive.google.com/file/d/1fC2BezUSsSQy7_H65A0ZfrYK0RB3TXXj/view?usp=sharing).
 
 <details>
@@ -175,7 +182,7 @@ Out[19]: 1684276
 `data['name']` is the image pair info, organized as [`scene_id`, `seq_id`, `image0_id`, `image1_id`].
 
 `data['score']` is the overlapping score defined in [SuperGlue](https://arxiv.org/pdf/1911.11763) (Page 12).
-</details>
+</details> -->
 
 
 ## Training

assets/megadepth_test_1500_scene_info/megadepth_test_1500.txt

@@ -1,5 +1,5 @@
-0022_0.1_0.3.npz
+0022_0.1_0.3
-0015_0.1_0.3.npz
+0015_0.1_0.3
-0015_0.3_0.5.npz
+0015_0.3_0.5
-0022_0.3_0.5.npz
+0022_0.3_0.5
-0022_0.5_0.7.npz
+0022_0.5_0.7

configs/data/megadepth_test_1500.py

@@ -3,9 +3,9 @@ from configs.data.base import cfg
 TEST_BASE_PATH = "assets/megadepth_test_1500_scene_info"
 
 cfg.DATASET.TEST_DATA_SOURCE = "MegaDepth"
-cfg.DATASET.TEST_DATA_ROOT = "/data/MegaDepth/megadepth_test_1500"
+cfg.DATASET.TEST_DATA_ROOT = "data/megadepth/test"
 cfg.DATASET.TEST_NPZ_ROOT = f"{TEST_BASE_PATH}"
 cfg.DATASET.TEST_LIST_PATH = f"{TEST_BASE_PATH}/megadepth_test_1500.txt"
 
 cfg.DATASET.MGDPT_IMG_RESIZE = 840
-cfg.DATASET.MIN_OVERLAP_SCORE = 0.0
+cfg.DATASET.MIN_OVERLAP_SCORE_TEST = 0.0

configs/data/scannet_test_1500.py

@@ -3,9 +3,9 @@ from configs.data.base import cfg
 TEST_BASE_PATH = "assets/scannet_test_1500"
 
 cfg.DATASET.TEST_DATA_SOURCE = "ScanNet"
-cfg.DATASET.TEST_DATA_ROOT = "/data/scannet/scannet_test_1500"
+cfg.DATASET.TEST_DATA_ROOT = "data/scannet/test"
 cfg.DATASET.TEST_NPZ_ROOT = f"{TEST_BASE_PATH}"
 cfg.DATASET.TEST_LIST_PATH = f"{TEST_BASE_PATH}/scannet_test.txt"
 cfg.DATASET.TEST_INTRINSIC_PATH = f"{TEST_BASE_PATH}/intrinsics.npz"
 
-cfg.DATASET.MIN_OVERLAP_SCORE = 0.0
+cfg.DATASET.MIN_OVERLAP_SCORE_TEST = 0.0

data/scannet/test

@@ -1 +0,0 @@
-/mnt/lustre/share/3dv/dataset/scannet/scannet_1500_testset

data/scannet/test/.gitignore

@@ -0,0 +1,3 @@
+*
+*/
+!.gitignore

docs/TRAINING.md

@@ -37,8 +37,8 @@ ln -s /path/to/scannet_indices/* /path/to/LoFTR/data/scannet/index
 
 # megadepth
 # -- # train and test dataset (train and test share the same dataset)
-ln -s /path/to/megadepth/Undistorted_SfM/* /path/to/LoFTR/data/megadepth/train
+ln -s /path/to/megadepth/Undistorted_SfM /path/to/LoFTR/data/megadepth/train
-ln -s /path/to/megadepth/Undistorted_SfM/* /path/to/LoFTR/data/megadepth/test
+ln -s /path/to/megadepth/Undistorted_SfM /path/to/LoFTR/data/megadepth/test
 # -- # dataset indices
 ln -s /path/to/megadepth_indices/* /path/to/LoFTR/data/megadepth/index
 ```

scripts/reproduce_test/indoor_ds.sh

@@ -8,7 +8,7 @@ export PYTHONPATH=$PROJECT_DIR:$PYTHONPATH
 cd $PROJECT_DIR
 
 data_cfg_path="configs/data/scannet_test_1500.py"
-main_cfg_path="configs/loftr/loftr_ds.py"
+main_cfg_path="configs/loftr/indoor/loftr_ds.py"
 ckpt_path="weights/indoor_ds.ckpt"
 dump_dir="dump/loftr_ds_indoor"
 profiler_name="inference"

scripts/reproduce_test/indoor_ot.sh

@@ -8,7 +8,7 @@ export PYTHONPATH=$PROJECT_DIR:$PYTHONPATH
 cd $PROJECT_DIR
 
 data_cfg_path="configs/data/scannet_test_1500.py"
-main_cfg_path="configs/loftr/loftr_ot.py"
+main_cfg_path="configs/loftr/indoor/loftr_ot.py"
 ckpt_path="weights/indoor_ot.ckpt"
 dump_dir="dump/loftr_ot_indoor"
 profiler_name="inference"

scripts/reproduce_test/outdoor_ds.sh

@@ -8,7 +8,7 @@ export PYTHONPATH=$PROJECT_DIR:$PYTHONPATH
 cd $PROJECT_DIR
 
 data_cfg_path="configs/data/megadepth_test_1500.py"
-main_cfg_path="configs/loftr/loftr_ds.py"
+main_cfg_path="configs/loftr/outdoor/loftr_ds.py"
 ckpt_path="weights/outdoor_ds.ckpt"
 dump_dir="dump/loftr_ds_outdoor"
 profiler_name="inference"

scripts/reproduce_test/outdoor_ot.sh

@@ -8,7 +8,7 @@ export PYTHONPATH=$PROJECT_DIR:$PYTHONPATH
 cd $PROJECT_DIR
 
 data_cfg_path="configs/data/megadepth_test_1500.py"
-main_cfg_path="configs/loftr/loftr_ot.py"
+main_cfg_path="configs/loftr/outdoor/loftr_ot.py"
 ckpt_path="weights/outdoor_ot.ckpt"
 dump_dir="dump/loftr_ot_outdoor"
 profiler_name="inference"

src/config/default.py

@@ -99,6 +99,7 @@ _CN.DATASET.MGDPT_DF = 8
 
 ##############  Trainer  ##############
 _CN.TRAINER = CN()
+_CN.TRAINER.WORLD_SIZE = 1
 _CN.TRAINER.CANONICAL_BS = 64
 _CN.TRAINER.CANONICAL_LR = 6e-3
 _CN.TRAINER.SCALING = None  # this will be calculated automatically

src/lightning/data.py

@@ -75,13 +75,13 @@ class MultiSceneDataModule(pl.LightningDataModule):
         self.train_loader_params = {
             'batch_size': args.batch_size,
             'num_workers': args.num_workers,
-            'pin_memory': args.pin_memory,
+            'pin_memory': getattr(args, 'pin_memory', True)
         }
         self.val_loader_params = {
             'batch_size': 1,
             'shuffle': False,
             'num_workers': args.num_workers,
-            'pin_memory': args.pin_memory,
+            'pin_memory': getattr(args, 'pin_memory', True)
         }
         self.test_loader_params = {
             'batch_size': 1,

src/utils/plotting.py

@@ -16,10 +16,6 @@ def _compute_conf_thresh(data):
 
 
 # --- VISUALIZATION --- #
-def plot_keypoints(axes, kpts0, kpts1, color='w', ps=2):
-    axes[0].scatter(kpts0[:, 0], kpts0[:, 1], c=color, s=ps)
-    axes[1].scatter(kpts1[:, 0], kpts1[:, 1], c=color, s=ps)
-
 
 def make_matching_figure(
         img0, img1, mkpts0, mkpts1, color,
@@ -38,8 +34,8 @@ def make_matching_figure(
     
     if kpts0 is not None:
         assert kpts1 is not None
-        # plot_keypoints(axes, kpts0, kpts1, color='k', ps=4)
+        axes[0].scatter(kpts0[:, 0], kpts0[:, 1], c='w', s=2)
-        plot_keypoints(axes, kpts0, kpts1, color='w', ps=2)
+        axes[1].scatter(kpts1[:, 0], kpts1[:, 1], c='w', s=2)
 
     # draw matches
     if mkpts0.shape[0] != 0 and mkpts1.shape[0] != 0:
@@ -98,7 +94,7 @@ def _make_evaluation_figure(data, b_id, alpha='dynamic'):
     color = error_colormap(epi_errs, conf_thr, alpha=alpha)
     
     text = [
-        f'Matches {len(kpts0)}',
+        f'#Matches {len(kpts0)}',
         f'Precision({conf_thr:.2e}) ({100 * precision:.1f}%): {n_correct}/{len(kpts0)}',
         f'Recall({conf_thr:.2e}) ({100 * recall:.1f}%): {n_correct}/{n_gt_matches}'
     ]
@@ -115,15 +111,12 @@ def _make_confidence_figure(data, b_id):
 
 def make_matching_figures(data, config, mode='evaluation'):
     """ Make matching figures for a batch.
+    
     Args:
         data (Dict): a batch updated by PL_LoFTR.
         config (Dict): matcher config
     Returns:
         figures (Dict[str, List[plt.figure]]
-    TODO:
-        - confidence mode plotting
-        - parallel plotting
-        - evaluation mode & confidence mode at the same time
     """
     assert mode in ['evaluation', 'confidence']  # 'confidence'
     figures = {mode: []}