Skip to content

zexianghui/Learning-Dense-Flow-Field-for-Highly-accurate-Cross-view-Camera-Localization

Repository files navigation

Dataset

KITTI: please refer to https://github.com/shiyujiao/HighlyAccurate to download the dataset. Your dataset folder structure should be like:

KITTI: raw_data: 2011_09_26: 2011_09_26_drive_0001_sync: image_00: image_01: image_02: image_03: calib_cam_to_cam.txt 2011_09_28: 2011_09_29: 2011_09_30: 2011_10_03: satmap: 2011_09_26: 2011_09_26_drive_0001_sync: 2011_09_29: 2011_09_30: 2011_10_03:

Ford multi-AV: please refer to https://github.com/shiyujiao/HighlyAccurate to download the dataset. Your dataset folder structure should be like:

Ford: 2017-08-04: V2: Log1: 2017-08-04-V2-Log1-FL: SatelliteMaps_18: grd_sat_quaternion_latlon.txt grd_sat_quaternion_latlon_test.txt 2017-10-26: Calibration-V2:

VIGOR: please refer to https://github.com/Jeff-Zilence/VIGOR and https://github.com/tudelft-iv/SliceMatch to download the dataset. Your dataset folder structure should be like:

VIGOR: Chicago: panorama: satellite: NewYork: SanFrancisco: Seattle: splits_corrected:

Oxford RobotCar: For instructions on how to obtain the dataset, please visit https://github.com/tudelft-iv/CrossViewMetricLocalization.

Codes

  1. Training:

python BEV_KITTI_train.py

python BEV_Ford_train.py --train_log_start 0 --train_log_end 1 python BEV_Ford_train.py --train_log_start 1 --train_log_end 2

python BEV_VIGOR_train.py --area cross --rotation_range 0 python BEV_VIGOR_train.py --area cross --rotation_range 180 python BEV_VIGOR_train.py --area same --rotation_range 0 python BEV_VIGOR_train.py --area same --rotation_range 180

python BEV_oxford_train.py

  1. Evaluation:

python BEV_KITTI_test.py

python BEV_Ford_test.py -- test_log_ind 0 python BEV_Ford_test.py -- test_log_ind 1

python BEV_VIGOR_test.py --area cross --rotation_range 0 python BEV_VIGOR_test.py --area cross --rotation_range 180 python BEV_VIGOR_test.py --area same --rotation_range 0 python BEV_VIGOR_test.py --area same --rotation_range 180

python BEV_oxford_test.py

Files to be downloaded

  1. Some files from the Oxford dataset

Some files from the Oxford dataset can be downloaded here and then placed in the dataLoader directory. https://drive.google.com/drive/folders/1B4RAqGwECydLgj4eeAVtnOxx8ob6sNBn?usp=drive_link

  1. Model files https://drive.google.com/drive/folders/1sqIATdj5U-v21DyW31hTUpX4el76LTeF?usp=drive_link

About

Learning Dense Flow Field for Highly-accurate Cross-view Camera Localization

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages