In accordance with the statutory mission and activities of ISPRS, the Society shall provide funds to support scientific and other initiatives, which will further improve its international status in the field of the photogrammetry, remote sensing and spatial information sciences, and will therefore benefit all ISPRS members. For 2017, the following seven projects were selected and approved by the Council for funding.
Title |
PI(s) |
TC/WG |
Benchmark database for tropical agricultural remote sensing application |
Ieda Del’Arco Sanches |
TC I |
SVSC – The UAV Semantic Video Segmentation Challenge 2017 |
Michael Ying Yang and Alper Yilmaz |
WGII/5 |
ISPRS Benchmark Challenge on Large Scale Classification of VHR Geospatial Data |
Ribana Roscher |
WGII/6 |
Benchmark for Multi-Platform Hyperspectral Image Processing |
Konstantinos Karantzalos |
WG III/4 |
ISPRS Benchmark on Indoor Modeling |
Kourosh Khoshelham |
WG IV/5 |
An ISPRS contribution to Transforming our World: Integration of GlobeLand30 with additional geospatial and socio-economic data for monitoring United Nation Sustainable Development Goals |
Jon Mills and Hao Wu |
ICWG IV/III |
Development of Educational Content: "Small UAS in Civil Engineering Applications" |
Roman Shults |
WG V/7 |
The following provides a brief summary of the above-awarded projects together with the information of their principle investigator(s) and co-investigator(s):
Benchmark database for tropical agricultural remote sensing application
Principle Investigator: Ieda Del'Arco Sanches, National Institute for Space Research, Brazil
Co-investigators: Raul Queiroz Feitosa, Pontifical Catholic University of Rio de Janeiro, Brazil; Alfredo José Barreto Luiz, Brazilian Agricultural Research Corporation, Brazil; Marinalva Dias Soares, Pedro Marco Achanccaray Diaz, Pontifical Catholic University of Rio de Janeiro, Brazil; Victor Hugo Rohden Prudente, Bruno Montibeller, National Institute for Space Research, Brazil
Food security is a major concern worldwide. In order to assure that the food production meets the world population demands at all times it is important to monitor agriculture activities at a regular basis. Such information can be derived from remote sensing data. In spite of topic’s relevance relatively few efforts have been invested to exploit modern pattern recognition and machine learning methods for agricultural land-cover mapping from multi-temporal, multi-sensor earth observation data. Furthermore, only a small proportion of the works published on this topic thus far relates to tropical/subtropical regions, where crop dynamics is more complicated and difficult to model than in temperate regions. A major hindrance has been the lack of accurate public databases for the comparison of different classification methods. The aim of the present project is to create and share a multi-temporal and multi-sensor benchmark database that can be used by the remote sensing community for agricultural land-cover mapping. The proposal is to collect information about crops in situ, in an important Brazilian agricultural area, to create field references maps including information about first and second harvests crops. In the same time, to acquire multi-temporal remote sensing images, from both active and passive orbital sensors, along the development of the main annual crops. The database (field data, reference maps and pre-processed images) can then be used as a base data for developing and testing image data processing and analysis methodologies, by different researchers, and the performance of the distinct approaches can be evaluated and compared in a true manner.
Final Report »
SVSC – The UAV Semantic Video Segmentation Challenge 2017
Principle Investigators: Michael Ying Yang, ITC, University of Twente, the Netherlands; Alper Yilmaz, PCV Lab, Ohio State University, USA
ISPRS Benchmark on UAV Semantic Video Segmentation (SVSB) is aiming at stimulating research and excellent work at the interface of two distinct disciplines working with geospatial data – Computer Vision and Photogrammetry communities. This is going to be achieved by acquiring very high-resolution video sequences from the UAV platform, which emerges as hot topics in both communities. The computer vision community has relied on several centralized benchmarks for performance evaluation of numerous tasks including object detection, pedestrian detection, 3D reconstruction, optical flow, tracking, and stereo estimation. Such benchmarks have proved to be extremely helpful to advance the state-of-the-art in the respective research fields. Interestingly, there has been rather limited benchmark effort on the ISPRS community, with exceptions of ISPRS benchmark on urban object detection and 3D building reconstruction, and benchmark dataset for Multi-Platform Very High Resolution Photogrammetry. With this benchmark, we would like to pave the way for a unified framework towards meaningful quantification of semantic segmentation from UAV videos. It is expected that SVSB will enhance the visibility of ISPRS research and events, attracting Computer Vision researchers joining ISPRS communities.
Final Report »
ISPRS Benchmark Challenge on Large Scale Classification of VHR Geospatial Data
Principle Investigator: Ribana Roscher, Institute of Geodesy and Geoinformation, University of Bonn, Germany
Remote sensing data offers the unique possibility to continuously monitor the Earth's changes over large areas. Nowadays, we are facing an unprecedented growth in the number of satellites with embedded sensors characterized by always increasing spatial, spectral and temporal resolutions. This drastically increases the amount of available data, particularly very high spatial resolution data (VHR), allowing both a high spatial and high temporal resolution coverage of the entire globe. However, the sheer size of this data bulk calls for new, automated methods for analysis, which will rely on machine learning as a workhorse.
In order to be capable of handling huge amounts of imagery data every day, these automated techniques shall be very efficient and applicable to large-scale scene analysis to reliably solve Earth monitoring tasks. A typical bottleneck of supervised learning approaches is the availability of (manually) labeled training data, which is particularly important to train state-of-the-art (deep) learning methods.
The ISPRS WG II/6 benchmark aims at addressing this issue by providing a publicly available, large-scale, VHR, multi-spectral dataset for training and evaluation of large-scale machine learning models. Interested researchers from various communities such as photogrammetry remote sensing, but also computer vision and machine learning will be able to evaluate their methods and algorithms on tasks such as object detection and semantic segmentation. Results can be submitted in the context of a benchmark challenge and will be compared with ground truth information using a sound, comprehensive and transparent evaluation tool.
Final Report »
Benchmark for Multi-platform Hyperspectral Image Processing
Principle Investigator: Konstantinos Karantzalos, National Technical University of Athens, Greece
Co-investigators: Eija Honkavaara, Finnish Geospatial Research Institute, Finland; Baoxin Hu, York University, Canada; Pölönen Ilkka, University of Jyväskylä, Finland; Zach. Kandylakis, Christos Kontopoulos, National Technical University of Athens, Greece
The ISPRS Scientific Initiative for a novel Benchmark on Multi-platform Hyperspectral Image Processing aims at promoting research and excellence on hyperspectral image processing through the efficient quantitative validation of state-of-the art techniques on hyperspectral datasets acquired from different satellite, airborne and UAVs acquisition platforms. The main goal is to fill the current gap regarding the limited availability of hyperspectral datasets and benchmarking frameworks for validating novel processing and data analytics methodologies against the state-of-the-art. To this end, the goal is to tackle these issues by: i) introducing novel, multi-platform hyperspectral imaging datasets as well as ii) introducing a benchmark framework for validating novel detection and classification algorithms. In particular, new datasets along with their corresponding ground truth will be gathered with different spatial and spectral resolution, acquired from different acquisition platforms. An online benchmarking framework will be also implemented for quantitatively comparing novel methodologies against the state-of-the-art.
Final Report »
ISPRS Benchmark on Indoor Modeling
Principle Investigator: Kourosh Khoshelham, University of Melbourne, Australia
Co-investigators: Lucía Díaz-Vilariño, University of Vigo, Spain; Michael Peter, University of Twente, Netherlands; Zhizhong Kang, China University of Geosciences, China
Up-to-date spatial models of indoor environments are needed in a growing number of applications, including navigation, emergency response and a range of location-based services. In recent years, several approaches to automated generation of indoor models from point clouds have been developed. However, until now it has not been possible to experimentally compare the performance of these methods due to the lack of benchmark datasets and a common evaluation framework. This ISPRS scientific initiative addresses this issue by creating a benchmark dataset comprising several point clouds captured by different sensors in various indoor environments. The project will also organise a benchmark test for the evaluation and comparison of indoor modelling methods based on manually created reference models and appropriate quality evaluation criteria. The datasets will be available for download from the ISPRS website, and interested participants will be invited to test their methods and submit their results for evaluation. The submitted models will be evaluated from a geometric, semantic and topologic point of view, and the results will be published on the ISPRS website.
Final Report »
An ISPRS contribution to Transforming Our World: augmentation of GlobeLand30 with additional data for monitoring United Nations Sustainable Development Goals
Principle Investigators: Jon Mills, Newcastle University, UK; Hao Wu, National Geomatics Center of China, China
Co-investigators: Sisi Zlatanova, Delft University of Technology, Delft; Klaus-Ulrich Komp, EFTAS Remote Sensing Transfer of Technology, Germany; Zhenhong Li, Newcastle University, UK; Zhilin Li, Hong Kong Polytechnic, Hong Kong; Miaole Hou, Beijing University of Civil Engineering and Architecture, China; Martin Herold, University of Wageningen, Netherlands
In September 2015, the United Nations (UN) adopted “Transforming Our World: the 2030 Agenda for Sustainable Development” as a new ambitious global development plan to end extreme poverty, fight inequality and injustice, and fix climate change. Almost exactly one year earlier, in September 2014, the release of the 30 m spatial resolution global land cover dataset, GlobeLand30, had heralded a step-change in the level of freely available geospatial information about our world. The principal aim of this ISPRS Scientific Initiatives proposal is therefore to assess the potential of GlobeLand30, when combined together with other geospatial and socio-economic datasets as deemed appropriate, to measure, monitor and manage the progress of sustainable development, as defined by the UN’s 17 Sustainable Development Goals. In order to achieve this aim, the project will firstly identify the Sustainable Development Indicators (SDIs) that can be quantified by GlobeLand30, independently or when augmented with additional earth observation (EO) datasets of relevance to the ISPRS community and appropriate socio-economic data. Once identified, the key scientific challenges and technology gaps behind each GlobeLand30 relevant SDI will be exposed in order to propose potential future solutions. Finally, approaches to visualise EO-derived SDIs will be investigated and effective communication will be demonstrated to end users and policy makers through real-world examples delivered via a prototype dynamic map atlas. The project will operate under the auspices of ISPRS ICWG IV/III, Global Mapping: Updating, Verification and Interoperability, with primary activities to include a 2017 workshop and an open competition to develop showcase SDI demonstrations.
Final Report »
Development of Educational Content: "Small UAS in Civil Engineering Applications"
Principle Investigators: Roman Shults, Kyiv National University of Construction and Architecture, Ukraine; Laxmi Thapa, Survey Department Ministry of Land Reform and Management Min Bhawan, Nepal
Co-investigators: Gottfried Konecny, Leibniz University Hannover, Germany; Eugene Levin, Michigan Technological University, USA; Vladimir A. Seredovich, Siberian State University of Architecture and Civil Engineering, Russian Federation; Karel Vach, EuroGV, Czech Republic
In today's world, decision support in multiple civil engineering application scenarios is widely supported by geospatial information. Challenge in training of modern civil engineers and architects is to provide them with balanced education and training in geospatial science and technology to prepare them for collection and processing of that data. To meet this challenge, ISPRS WG V/7 proposes to develop educational content “Small UAS in Civil Engineering Application Scenarios” (SUAS-CAS). SUAS-CAS is aimed in development of the unique computer based educational materials that will optimally integrate education in fundamental problems associated with geospatial data obtaining and processing in civil engineering applications with training in SUAS systems use in response to that kind of applications.
Due to commonly accepted research and development projects methodology, SUAS-CAS will be realized towards following project tasks:
- Analytical review and selection of rotary and fixed wing SUAS platforms suitable for the project.
- Analytical review and selection of commercial-off-the-shelf and open source software systems.
- Preparation of the SUAS-CAS syllabus.
- Preparation and multimedia recording of the SUAS-CAS lectures.
- Establishing of SUAS-CAS hands-on components.
- Develop SUAS-CAS web-portal.
- SUAS-CAS approbation at ISPRS Kyiv workshop.
- Create SUAS-CAS outreach and data dissemination strategy.
SUAS-CAS project will be culminated in establishing and approbation of the educational modules, which will include online training manual on the collection and processing of UAV geospatial data, educational website available for training and education of civil engineers and architects worldwide.
Final Report »