Resolve merge conflict

.github/scripts/download_single_benchmark.sh

@@ -71,6 +71,11 @@ function download_and_unzip_dataset_files {
     # Tanks and Temples Dataset, "Barn" scene.
     WGET_URL1=https://github.com/johnwlambert/gtsfm-datasets-mirror/releases/download/tanks-and-temples-barn/Tanks_and_Temples_Barn_410.zip
     ZIP_FNAME=Tanks_and_Temples_Barn_410.zip
+
+  elif [ "$DATASET_NAME" == "south-building-128" ]; then
+    WGET_URL1=https://github.com/johnwlambert/gtsfm-datasets-mirror/releases/download/south-building-128/south-building-128.zip
+    ZIP_FNAME=south-building-128.zip
+
   fi
 
   # Download the data.
@@ -139,6 +144,9 @@ function download_and_unzip_dataset_files {
     unzip gerrard-hall-100.zip
     ls -ltrh gerrard-hall-100
 
+  elif [ "$DATASET_NAME" == "south-building-128" ]; then
+    unzip south-building-128.zip
+
   elif [ "$DATASET_NAME" == "tanks-and-temples-barn-410" ]; then
     unzip -qq Tanks_and_Temples_Barn_410.zip
   fi

.github/scripts/execute_single_benchmark.sh

@@ -3,10 +3,9 @@
 DATASET_NAME=$1
 CONFIG_NAME=$2
 MAX_FRAME_LOOKAHEAD=$3
-IMAGE_EXTENSION=$4
-LOADER_NAME=$5
-MAX_RESOLUTION=$6
-SHARE_INTRINSICS=$7
+LOADER_NAME=$4
+MAX_RESOLUTION=$5
+SHARE_INTRINSICS=$6
 
 # Extract the data, configure arguments for runner.
 if [ "$DATASET_NAME" == "door-12" ]; then
@@ -33,10 +32,14 @@ elif [ "$DATASET_NAME" == "gerrard-hall-100" ]; then
 elif [ "$DATASET_NAME" == "tanks-and-temples-barn-410" ]; then
   DATASET_ROOT="Tanks_and_Temples_Barn_410"
   SCENE_NAME="Barn"
+elif [ "$DATASET_NAME" == "south-building-128" ]; then
+  IMAGES_DIR=south-building-128/images
+  #COLMAP_FILES_DIRPATH=south-building-128/colmap-official-2016-10-05
+  COLMAP_FILES_DIRPATH=south-building-128/colmap-2023-07-28-txt
 fi
 
 echo "Config: ${CONFIG_NAME}, Loader: ${LOADER_NAME}"
-echo "Max. Frame Lookahead: ${MAX_FRAME_LOOKAHEAD}, Image Extension: ${IMAGE_EXTENSION}, Max. Resolution: ${MAX_RESOLUTION}"
+echo "Max. Frame Lookahead: ${MAX_FRAME_LOOKAHEAD}, Max. Resolution: ${MAX_RESOLUTION}"
 echo "Share intrinsics for all images? ${SHARE_INTRINSICS}"
 
 # Setup the command line arg if intrinsics are to be shared
@@ -50,7 +53,8 @@ fi
 if [ "$LOADER_NAME" == "olsson-loader" ]; then
   python gtsfm/runner/run_scene_optimizer_olssonloader.py \
     --dataset_root $DATASET_ROOT \
-    --config_name ${CONFIG_NAME}.yaml \
+    --config_name unified \
+    --correspondence_generator_config_name ${CONFIG_NAME} \
     --max_frame_lookahead $MAX_FRAME_LOOKAHEAD \
     --max_resolution ${MAX_RESOLUTION} \
     ${SHARE_INTRINSICS_ARG} \
@@ -60,7 +64,8 @@ elif [ "$LOADER_NAME" == "colmap-loader" ]; then
   python gtsfm/runner/run_scene_optimizer_colmaploader.py \
     --images_dir ${IMAGES_DIR} \
     --colmap_files_dirpath $COLMAP_FILES_DIRPATH \
-    --config_name ${CONFIG_NAME}.yaml \
+    --config_name unified \
+    --correspondence_generator_config_name ${CONFIG_NAME} \
     --max_frame_lookahead $MAX_FRAME_LOOKAHEAD \
     --max_resolution ${MAX_RESOLUTION} \
     ${SHARE_INTRINSICS_ARG} \
@@ -69,7 +74,8 @@ elif [ "$LOADER_NAME" == "colmap-loader" ]; then
 elif [ "$LOADER_NAME" == "astrovision" ]; then
   python gtsfm/runner/run_scene_optimizer_astrovision.py \
     --data_dir $DATASET_ROOT \
-    --config_name ${CONFIG_NAME}.yaml \
+    --config_name unified \
+    --correspondence_generator_config_name ${CONFIG_NAME} \
     --max_frame_lookahead $MAX_FRAME_LOOKAHEAD \
     --max_resolution ${MAX_RESOLUTION} \
     ${SHARE_INTRINSICS_ARG} \

.github/scripts/execute_single_benchmark_self_hosted.sh

@@ -8,10 +8,9 @@ DATASET_PREFIX=/usr/local/gtsfm-data
 DATASET_NAME=$1
 CONFIG_NAME=$2
 MAX_FRAME_LOOKAHEAD=$3
-IMAGE_EXTENSION=$4
-LOADER_NAME=$5
-MAX_RESOLUTION=$6
-SHARE_INTRINSICS=$7
+LOADER_NAME=$4
+MAX_RESOLUTION=$5
+SHARE_INTRINSICS=$6
 
 # Extract the data, configure arguments for runner.
 if [ "$DATASET_NAME" == "skydio-501" ]; then
@@ -26,7 +25,7 @@ fi
 
 
 echo "Config: ${CONFIG_NAME}, Loader: ${LOADER_NAME}"
-echo "Max. Frame Lookahead: ${MAX_FRAME_LOOKAHEAD}, Image Extension: ${IMAGE_EXTENSION}, Max. Resolution: ${MAX_RESOLUTION}"
+echo "Max. Frame Lookahead: ${MAX_FRAME_LOOKAHEAD}, Max. Resolution: ${MAX_RESOLUTION}"
 echo "Share intrinsics for all images? ${SHARE_INTRINSICS}"
 
 # Setup the command line arg if intrinsics are to be shared

.github/workflows/benchmark-self-hosted.yml

@@ -30,7 +30,7 @@ jobs:
       PYTHON_VERSION: 3.8
 
     steps:
-      - uses: actions/checkout@v3.5.3
+      - uses: actions/checkout@v4.0.0
       - name: Cache frontend
         uses: actions/cache@v3
         env:
@@ -68,7 +68,6 @@ jobs:
           DATASET_NAME=${{ matrix.config_dataset_info[1] }}
           CONFIG_NAME=${{ matrix.config_dataset_info[0] }}
           MAX_FRAME_LOOKAHEAD=${{ matrix.config_dataset_info[2] }}
-          IMAGE_EXTENSION=${{ matrix.config_dataset_info[3] }}
           LOADER_NAME=${{ matrix.config_dataset_info[5] }}
           MAX_RESOLUTION=${{ matrix.config_dataset_info[6] }}
           SHARE_INTRINSICS=${{ matrix.config_dataset_info[7] }}
@@ -79,7 +78,6 @@ jobs:
             $DATASET_NAME \
             $CONFIG_NAME \
             $MAX_FRAME_LOOKAHEAD \
-            $IMAGE_EXTENSION \
             $LOADER_NAME \
             $MAX_RESOLUTION \
             $SHARE_INTRINSICS

.github/workflows/benchmark.yml

@@ -12,20 +12,21 @@ jobs:
       matrix:
         config_dataset_info:
           [
-            # config                     dataset                lookahead   img-extension  source      loader          max-res  share-intrinsics
-            [sift_front_end,             door-12,               12,         JPG,           test_data,  olsson-loader,  1296,    true],
-            [deep_front_end,             door-12,               12,         JPG,           test_data,  olsson-loader,  1296,    true],
-            [sift_front_end,             skydio-8,              8,          jpg,           gdrive ,    colmap-loader,  760,     true],
-            [deep_front_end,             skydio-8,              8,          jpg,           gdrive,     colmap-loader,  760,     true],
-            [sift_front_end,             skydio-32,             32,         jpg,           gdrive,     colmap-loader,  760,     true],
-            [deep_front_end,             skydio-32,             32,         jpg,           gdrive,     colmap-loader,  760,     true],
-            [sift_front_end,             palace-fine-arts-281,  25,         jpg,           wget,       olsson-loader,  320,     true],
-            [deep_front_end,             notre-dame-20,         20,         jpg,           gdrive,     colmap-loader,  760,     false],
-            [sift_front_end_astrovision, 2011205_rc3,           65,         png,           wget,       astrovision,    1024,    true],
-            [deep_front_end_astrovision, 2011205_rc3,           65,         png,           wget,       astrovision,    1024,    true],
-            [sift_front_end,             gerrard-hall-100,      0,          jpg,           wget,       colmap-loader,  760,     true],
-            [deep_front_end,             gerrard-hall-100,      0,          jpg,           wget,       colmap-loader,  760,     true],
-            [synthetic_front_end,   tanks-and-temples-barn-410, 4,          jpg,           wget,    tanks-and-temples, 1080,    true],
+            # config              dataset                lookahead   img-extension  source      loader          max-res  share-intrinsics
+            [sift,                door-12,               15,         JPG,           test_data,  olsson-loader,  1296,    true],
+            [lightglue,           door-12,               15,         JPG,           test_data,  olsson-loader,  1296,    true],
+            [sift,                skydio-8,              15,         jpg,           gdrive ,    colmap-loader,  760,     true],
+            [lightglue,           skydio-8,              15,         jpg,           gdrive,     colmap-loader,  760,     true],
+            [sift,                skydio-32,             15,         jpg,           gdrive,     colmap-loader,  760,     true],
+            [lightglue,           skydio-32,             15,         jpg,           gdrive,     colmap-loader,  760,     true],
+            [sift,                palace-fine-arts-281,  15,         jpg,           wget,       olsson-loader,  320,     true],
+            [lightglue,           notre-dame-20,         15,         jpg,           gdrive,     colmap-loader,  760,     false],
+            [sift,                2011205_rc3,           15,         png,           wget,       astrovision,    1024,    true],
+            [lightglue,           2011205_rc3,           15,         png,           wget,       astrovision,    1024,    true],
+            [sift,                gerrard-hall-100,      15,         jpg,           wget,       colmap-loader,  760,     true],
+            [lightglue,           gerrard-hall-100,      15,         jpg,           wget,       colmap-loader,  760,     true],
+            [sift,                south-building-128,    15,         jpg,           wget,       colmap-loader,  760,     true],
+            [lightglue,           south-building-128,    15,         jpg,           wget,       colmap-loader,  760,     true],  
           ]
     defaults:
       run:
@@ -35,7 +36,7 @@ jobs:
       PYTHON_VERSION: 3.8
 
     steps:
-      - uses: actions/checkout@v3.5.3
+      - uses: actions/checkout@v4.0.0
       - name: Cache frontend
         uses: actions/cache@v3
         env:
@@ -64,15 +65,13 @@ jobs:
           DATASET_NAME=${{ matrix.config_dataset_info[1] }}
           CONFIG_NAME=${{ matrix.config_dataset_info[0] }}
           MAX_FRAME_LOOKAHEAD=${{ matrix.config_dataset_info[2] }}
-          IMAGE_EXTENSION=${{ matrix.config_dataset_info[3] }}
           LOADER_NAME=${{ matrix.config_dataset_info[5] }}
           MAX_RESOLUTION=${{ matrix.config_dataset_info[6] }}
           SHARE_INTRINSICS=${{ matrix.config_dataset_info[7] }}
           bash .github/scripts/execute_single_benchmark.sh \
             $DATASET_NAME \
             $CONFIG_NAME \
             $MAX_FRAME_LOOKAHEAD \
-            $IMAGE_EXTENSION \
             $LOADER_NAME \
             $MAX_RESOLUTION \
             $SHARE_INTRINSICS

.github/workflows/build-wheels.yml

@@ -18,7 +18,7 @@ jobs:
         shell: bash -l {0}
 
     steps:
-      - uses: actions/checkout@v3.5.3
+      - uses: actions/checkout@v4.0.0
       - uses: conda-incubator/setup-miniconda@v2
         with:
           mamba-version: "*"

.github/workflows/test-python.yml

@@ -14,7 +14,7 @@ jobs:
       PYTHON_VERSION: 3.8
 
     steps:
-      - uses: actions/checkout@v3.5.3
+      - uses: actions/checkout@v4.0.0
       - uses: conda-incubator/setup-miniconda@v2
         with:
           mamba-version: "*"

.github/workflows/test-reproducibility.yml

@@ -14,7 +14,7 @@ jobs:
       PYTHON_VERSION: 3.8
 
     steps:
-      - uses: actions/checkout@v3.5.3
+      - uses: actions/checkout@v4.0.0
       - uses: conda-incubator/setup-miniconda@v2
         with:
           mamba-version: "*"

README.md

@@ -69,13 +69,13 @@ To run SfM with a dataset with only an image directory and EXIF, with image file
 and run
 
 ```python
-python gtsfm/runner/run_scene_optimizer_olssonloader.py --config_name {CONFIG_NAME} --dataset_root {DATASET_ROOT} --image_extension jpg --num_workers {NUM_WORKERS}
+python gtsfm/runner/run_scene_optimizer_olssonloader.py --config_name {CONFIG_NAME} --dataset_root {DATASET_ROOT} --num_workers {NUM_WORKERS}
 ```
 
 For example, if you had 4 cores available and wanted to use the Deep Front-End (recommended) on the "door" dataset, you should run:
 
 ```bash
-python gtsfm/runner/run_scene_optimizer_olssonloader.py --dataset_root tests/data/set1_lund_door --image_extension JPG --config_name deep_front_end.yaml --num_workers 4
+python gtsfm/runner/run_scene_optimizer_olssonloader.py --dataset_root tests/data/set1_lund_door --config_name deep_front_end.yaml --num_workers 4
 ```
 
 (or however many workers you desire).
@@ -87,7 +87,7 @@ Currently we require EXIF data embedded into your images (or you can provide gro
 If you would like to compare GTSfM output with COLMAP output, please run:
 
 ```python
-python gtsfm/runner/run_scene_optimizer_colmaploader.py --config_name {CONFIG_NAME} --images_dir {IMAGES_DIR} --colmap_files_dirpath {COLMAP_FILES_DIRPATH} --image_extension jpg --num_workers {NUM_WORKERS} --max_frame_lookahead {MAX_FRAME_LOOKAHEAD}
+python gtsfm/runner/run_scene_optimizer_colmaploader.py --config_name {CONFIG_NAME} --images_dir {IMAGES_DIR} --colmap_files_dirpath {COLMAP_FILES_DIRPATH} --num_workers {NUM_WORKERS} --max_frame_lookahead {MAX_FRAME_LOOKAHEAD}
 ```
 
 where `COLMAP_FILES_DIRPATH` is a directory where .txt files such as `cameras.txt`, `images.txt`, etc have been saved.
@@ -145,7 +145,3 @@ Open-source Python implementation:
 ```
 
 Note: authors are listed in alphabetical order (by last name).
-
-## Compiling Additional Verifiers
-
-On Linux, we have made `pycolmap`'s LORANSAC available in [pypi](https://pypi.org/project/pycolmap/). However, on Mac, `pycolmap` must be built from scratch. See the instructions [here](https://github.com/borglab/gtsfm/blob/master/gtsfm/frontend/verifier/loransac.py#L10).

environment_linux.yml

@@ -52,5 +52,5 @@ dependencies:
       - colour
       - pycolmap>=0.1.0
       - trimesh[easy]
-      - gtsam==4.2a8
+      - gtsam==4.2
       - pydot

environment_linux_cpuonly.yml

@@ -52,5 +52,5 @@ dependencies:
       - colour
       - pycolmap>=0.1.0
       - trimesh[easy]
-      - gtsam==4.2a8
+      - gtsam==4.2
       - pydot

environment_mac.yml

@@ -57,5 +57,5 @@ dependencies:
       - colour
       - trimesh[easy]
       - pycolmap>=0.1.0
-      - gtsam==4.2a8
+      - gtsam==4.2
       - pydot

gtsfm/averaging/rotation/rotation_averaging_base.py

@@ -3,6 +3,7 @@
 Authors: Jing Wu, Ayush Baid
 """
 import abc
+import time
 from typing import Dict, List, Optional, Tuple
 
 import dask
@@ -54,14 +55,45 @@ def run_rotation_averaging(
                 underconstrained system or ill-constrained system).
         """
 
+    def _run_rotation_averaging_base(
+        self,
+        num_images: int,
+        i2Ri1_dict: Dict[Tuple[int, int], Optional[Rot3]],
+        i1Ti2_priors: Dict[Tuple[int, int], PosePrior],
+        wTi_gt: List[Optional[Pose3]],
+    ) -> Tuple[List[Optional[Rot3]], GtsfmMetricsGroup]:
+        """Runs rotation averaging and computes metrics.
+
+        Args:
+            num_images: Number of poses.
+            i2Ri1_dict: Relative rotations as dictionary (i1, i2): i2Ri1.
+            i1Ti2_priors: Priors on relative poses as dictionary(i1, i2): PosePrior on i1Ti2.
+            wTi_gt: Ground truth global rotations to compare against.
+
+        Returns:
+            Global rotations for each camera pose, i.e. wRi, as a list. The number of entries in the list is
+                `num_images`. The list may contain `None` where the global rotation could not be computed (either
+                underconstrained system or ill-constrained system).
+            Metrics on global rotations.
+        """
+        start_time = time.time()
+        wRis = self.run_rotation_averaging(num_images, i2Ri1_dict, i1Ti2_priors)
+        run_time = time.time() - start_time
+
+        metrics = self.evaluate(wRis, wTi_gt)
+        metrics.add_metric(GtsfmMetric("total_duration_sec", run_time))
+
+        return wRis, metrics
+
     def evaluate(self, wRi_computed: List[Optional[Rot3]], wTi_gt: List[Optional[Pose3]]) -> GtsfmMetricsGroup:
-        """Evaluate the global rotations computed by the rotation averaging implementation.
+        """Evaluates the global rotations computed by the rotation averaging implementation.
 
         Args:
-            wRi_computed: list of global rotations computed.
-            wTi_gt: ground truth global rotations to compare against.
+            wRi_computed: List of global rotations computed.
+            wTi_gt: Ground truth global rotations to compare against.
+
         Raises:
-            ValueError: if the length of the computed and GT list differ.
+            ValueError: If the length of the computed and GT list differ.
 
         Returns:
             Metrics on global rotations.
@@ -97,7 +129,8 @@ def create_computation_graph(
             global rotations wrapped using dask.delayed.
         """
 
-        wRis = dask.delayed(self.run_rotation_averaging)(num_images, i2Ri1_graph, i1Ti2_priors)
-        metrics = dask.delayed(self.evaluate)(wRis, gt_wTi_list)
+        wRis, metrics = dask.delayed(self._run_rotation_averaging_base, nout=2)(
+            num_images, i2Ri1_dict=i2Ri1_graph, i1Ti2_priors=i1Ti2_priors, wTi_gt=gt_wTi_list
+        )
 
         return wRis, metrics

gtsfm/averaging/translation/averaging_1dsfm.py

@@ -10,9 +10,11 @@
 
 Authors: Jing Wu, Ayush Baid, Akshay Krishnan
 """
+import time
 from collections import defaultdict
 from enum import Enum
 from typing import DefaultDict, Dict, List, Optional, Set, Tuple
+
 import gtsam
 import numpy as np
 from gtsam import (
@@ -461,6 +463,7 @@ def run_translation_averaging(
 
         w_i2Ui1_dict, valid_cameras = get_valid_measurements_in_world_frame(i2Ui1_dict, wRi_list)
 
+        start_time = time.time()
         if self._use_tracks_for_averaging:
             if tracks_2d is None:
                 logger.info("No tracks provided for translation averaging. Falling back to camera unit translations.")
@@ -473,10 +476,13 @@ def run_translation_averaging(
         else:
             w_i2Ui1_dict_tracks = {}
 
+        inlier_computation_start_time = time.time()
         w_i2Ui1_dict_inliers, w_i2Ui1_dict_tracks_inliers, inlier_cameras = self.compute_inliers(
             w_i2Ui1_dict, w_i2Ui1_dict_tracks
         )
+        inlier_computation_time = time.time() - inlier_computation_start_time
 
+        averaging_start_time = time.time()
         wti_list = self.__run_averaging(
             num_images=num_images,
             w_i2Ui1_dict=w_i2Ui1_dict_inliers,
@@ -486,6 +492,7 @@ def run_translation_averaging(
             absolute_pose_priors=absolute_pose_priors,
             scale_factor=scale_factor,
         )
+        averaging_time = time.time() - averaging_start_time
 
         # Compute the metrics.
         ta_metrics = compute_metrics(set(w_i2Ui1_dict_inliers.keys()), i2Ui1_dict, wRi_list, wti_list, gt_wTi_list)
@@ -497,6 +504,12 @@ def run_translation_averaging(
         wTi_list = [
             Pose3(wRi, wti) if wRi is not None and wti is not None else None for wRi, wti in zip(wRi_list, wti_list)
         ]
+        total_time = time.time() - start_time
+        logger.info("Translation averaging took %.4f seconds.", total_time)
+        ta_metrics.add_metric(GtsfmMetric("total_duration_sec", total_time))
+        ta_metrics.add_metric(GtsfmMetric("outier_rejection_duration_sec", inlier_computation_time))
+        ta_metrics.add_metric(GtsfmMetric("optimization_duration_sec", averaging_time))
+
         return wTi_list, ta_metrics