Code

Implemented / contributed to the following repositories: Scroll to Datasets
Skaret View
aerial_renderer

A light-weight rendering program (openGL) that, given a geo-referenced orthoimage, elevation map, camera pose, camera and distortion model, is able to render a depth, optical, or semantic image that can be used for training data generation or online map tracking.
multipoint

MultiPoint: Feature detector and descriptor for optical/thermal image alignment.
aerial_mapper

A library containing basic code to generate geo-referenced dense point clouds, digital elevation models, and orthoimages.
robust_point_cloud_registration

A library containing the source code for point cloud registration using iterative probabilistic data association method. It contains wrappers for other point cloud registration methods.
Skaret View
aslam_cv2

A library containing commonly used functions for computer vision algorithms and data containers such as camera or distortion models.
maplab

A framework for visual-inertial mapping and localization.
kalibr

A toolbox for the intrinsic and extrinsic calibration of a or multiple cameras, the spatial-temporal calibration of a camera-IMU setup, and IMU-IMU calibration.
Pose-Graph Optimization

Python implementation of "A Tutorial on Graph-Based SLAM", (Alg. 1).
Standard ICP

Python implementation of standard ICP.

Datasets

Scroll to Code
Optical-Thermal-Infrared Human Detection Datasets (released soon)
The datasets have been collected by a fixed-wing and a rotary-wing UAV in realistic search-and-rescue scenarios. All datasets are annotated with bounding boxes.
Multispectral (Optical-Thermal) Image Pair Dataset
The data was collected using a fixed-wing UAV equipped with two downward facing cameras in ten flights at different times of the day. The RGB camera is a UI- 5261SE Rev. 4 with a 16 mm focal length lens and the thermal camera a FLIR TAU2 19 mm, spectral band 7.5–13.5 μm
Vision & Laser Datasets From A Heterogeneous UAV Fleet
Datasets consisting of optical and laser point clouds have been released for the robust_point_cloud_registration repository.
Aerial images and poses (synthetic and real-world)
As part of the aerial_mapper repository, we released synthetic and real-world images with corresponding ground-truth and estimated camera poses, camera intrinsics, and distortion parameters to encourage the development of new algorithms and to reproduce the results.
AtlantikSolar 81-hour solar-powered flight dataset
This is the dataset recorded from July 14th - 17th 2015 during AtlantikSolar's continuous solar-powered flight of 81-hours total duration.

For the case that a link becomes unavailable, please consult ASL's dataset website, ASL's code base, or write me an e-mail.