Adapt to the new pycolmap interface

hloc/localize_inloc.py

@@ -110,7 +110,7 @@ def pose_from_cluster(dataset_dir, q, retrieved, feature_file, match_file, skip=
         "height": height,
         "params": [focal_length, cx, cy],
     }
-    ret = pycolmap.absolute_pose_estimation(all_mkpq, all_mkp3d, cfg, 48.00)
+    ret = pycolmap.estimate_and_refine_absolute_pose(all_mkpq, all_mkp3d, cfg, 48.00)
     ret["cfg"] = cfg
     return ret, all_mkpq, all_mkpr, all_mkp3d, all_indices, num_matches
 

hloc/localize_sfm.py

@@ -58,7 +58,7 @@ def __init__(self, reconstruction, config=None):
     def localize(self, points2D_all, points2D_idxs, points3D_id, query_camera):
         points2D = points2D_all[points2D_idxs]
         points3D = [self.reconstruction.points3D[j].xyz for j in points3D_id]
-        ret = pycolmap.absolute_pose_estimation(
+        ret = pycolmap.estimate_and_refine_absolute_pose(
             points2D,
             points3D,
             query_camera,

hloc/pipelines/7Scenes/create_gt_sfm.py

@@ -11,7 +11,16 @@
 
 def scene_coordinates(p2D, R_w2c, t_w2c, depth, camera):
     assert len(depth) == len(p2D)
-    p2D_norm = np.stack(pycolmap.Camera(camera._asdict()).image_to_world(p2D))
+    pycolmap_camera = pycolmap.Camera(
+        {
+            "camera_id": camera.id,
+            "model": camera.model,
+            "width": camera.width,
+            "height": camera.height,
+            "params": camera.params,
+        }
+    )
+    p2D_norm = pycolmap_camera.cam_from_img(p2D)
     p2D_h = np.concatenate([p2D_norm, np.ones_like(p2D_norm[:, :1])], 1)
     p3D_c = p2D_h * depth[:, None]
     p3D_w = (p3D_c - t_w2c) @ R_w2c
@@ -52,7 +61,16 @@ def project_to_image(p3D, R, t, camera, eps: float = 1e-4, pad: int = 1):
     p3D = (p3D @ R.T) + t
     visible = p3D[:, -1] >= eps  # keep points in front of the camera
     p2D_norm = p3D[:, :-1] / p3D[:, -1:].clip(min=eps)
-    p2D = np.stack(pycolmap.Camera(camera._asdict()).world_to_image(p2D_norm))
+    pycolmap_camera = pycolmap.Camera(
+        {
+            "camera_id": camera.id,
+            "model": camera.model,
+            "width": camera.width,
+            "height": camera.height,
+            "params": camera.params,
+        }
+    )
+    p2D = pycolmap_camera.img_from_cam(p2D_norm)
     size = np.array([camera.width - pad - 1, camera.height - pad - 1])
     valid = np.all((p2D >= pad) & (p2D <= size), -1)
     valid &= visible

hloc/utils/geometry.py

@@ -7,7 +7,7 @@ def to_homogeneous(p):
 
 
 def compute_epipolar_errors(j_from_i: pycolmap.Rigid3d, p2d_i, p2d_j):
-    j_E_i = j_from_i.essential_matrix()
+    j_E_i = pycolmap.essential_matrix_from_pose(j_from_i)
     l2d_j = to_homogeneous(p2d_i) @ j_E_i.T
     l2d_i = to_homogeneous(p2d_j) @ j_E_i
     dist = np.abs(np.sum(to_homogeneous(p2d_i) * l2d_i, axis=1))

hloc/visualization.py

@@ -17,7 +17,7 @@ def visualize_sfm_2d(
         reconstruction = pycolmap.Reconstruction(reconstruction)
 
     if not selected:
-        image_ids = reconstruction.reg_image_ids()
+        image_ids = list(reconstruction.reg_image_ids())
         selected = random.Random(seed).sample(image_ids, min(n, len(image_ids)))
 
     for i in selected:
@@ -110,7 +110,7 @@ def visualize_loc_from_log(
         # select the first, largest cluster if the localization failed
         loc = loc["log_clusters"][loc["best_cluster"] or 0]
 
-    inliers = np.array(loc["PnP_ret"]["inliers"])
+    inliers = np.array(loc["PnP_ret"]["inlier_mask"])
     mkp_q = loc["keypoints_query"]
     n = len(loc["db"])
     if reconstruction is not None:

requirements.txt

@@ -7,7 +7,7 @@ matplotlib
 plotly
 scipy
 h5py
-pycolmap>=0.6.0
+pycolmap>=3.11.1
 kornia>=0.6.11
 gdown
 lightglue @ git+https://github.com/cvg/LightGlue