This is an implementation for TensorFlow of the PoseNet architecture
As described in the ICCV 2015 paper PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization Alex Kendall, Matthew Grimes and Roberto Cipolla [http://mi.eng.cam.ac.uk/projects/relocalisation/]
-
Download the Cambridge Landmarks King's College dataset from here.
-
Download the starting weights from here.
-
The PoseNet model is defined in the posenet.py file
-
The starting weights (posenet.npy) for training were obtained by converting caffemodel weights from here.
-
To run:
- Extract the King's College dataset to wherever you prefer
- Extract the starting weights
- To train, simply run train.py after setting the path to the King's College dataset (note this will take a long time)
- After training, to test run test.py after updating the paths on these lines