Merge pull request #1211 from ducha-aiki/pycolmap

Add pycolmap

Author: djherbis

Dockerfile.tmpl

@@ -36,7 +36,7 @@ ARG TORCHAUDIO_VERSION
 ARG TORCHTEXT_VERSION
 ARG TORCHVISION_VERSION
 
-# Disable pesky logs like: KMP_AFFINITY: pid 6121 tid 6121 thread 0 bound to OS proc set 0 
+# Disable pesky logs like: KMP_AFFINITY: pid 6121 tid 6121 thread 0 bound to OS proc set 0
 # See: https://stackoverflow.com/questions/57385766/disable-tensorflow-log-information
 ENV KMP_WARNINGS=0
 # Also make the KMP logs noverbose.
@@ -221,7 +221,7 @@ RUN pip install ibis-framework && \
 
 RUN pip install scipy && \
     pip install scikit-learn && \
-    # Scikit-learn accelerated library for x86 
+    # Scikit-learn accelerated library for x86
     # TODO(b/262387811#4) Unpin when the package is fixed.
     pip install scikit-learn-intelex==2021.6.3 && \
     # HDF5 support
@@ -336,7 +336,7 @@ RUN pip install mpld3 && \
     pip install mock && \
     /tmp/clean-layer.sh
 
-RUN pip install tensorpack && \   
+RUN pip install tensorpack && \
     # Add google PAIR-code Facets
     cd /opt/ && git clone https://github.com/PAIR-code/facets && cd facets/ && jupyter nbextension install facets-dist/ --user && \
     export PYTHONPATH=$PYTHONPATH:/opt/facets/facets_overview/python/ && \
@@ -374,7 +374,7 @@ RUN pip install --upgrade cython && \
     # b/183041606#comment5: the Kaggle data proxy doesn't support these APIs. If the library is missing, it falls back to using a regular BigQuery query to fetch data.
     pip uninstall -y google-cloud-bigquery-storage && \
     # After launch this should be installed from pip
-    pip install git+https://github.com/googleapis/python-aiplatform.git@mb-release && \ 
+    pip install git+https://github.com/googleapis/python-aiplatform.git@mb-release && \
     pip install ortools && \
     pip install scattertext && \
     # Pandas data reader
@@ -493,7 +493,7 @@ RUN pip install flashtext && \
     # b/254245259 catalyst requires accelerate but it breaks with the version 0.13.1
     pip install accelerate==0.12.0 && \
     pip install catalyst && \
-    # b/206990323 osmx 1.1.2 requires numpy >= 1.21 which we don't want. 
+    # b/206990323 osmx 1.1.2 requires numpy >= 1.21 which we don't want.
     pip install osmnx==1.1.1 && \
     apt-get -y install libspatialindex-dev && \
     pip install pytorch-ignite && \
@@ -526,6 +526,7 @@ RUN pip install flashtext && \
     pip install tables && \
     pip install openpyxl && \
     pip install timm && \
+    pip install pycolmap && \
     /tmp/clean-layer.sh
 
 # Download base easyocr models.

tests/test_pycolmap.py

@@ -0,0 +1,13 @@
+import unittest
+
+import numpy as np
+import cv2
+import pycolmap
+
+class TestPycolmap(unittest.TestCase):
+    def test_sift(self):
+        img = cv2.cvtColor(cv2.imread('/input/tests/data/face.jpg'), cv2.COLOR_BGR2GRAY)
+        img = img.astype(np.float32)/255.
+        sift = pycolmap.Sift()
+        keypoints, scores, descriptors = sift.extract(img)
+        self.assertEqual(keypoints.shape, (64, 4))