Groupwise Registration of Aerial Images

Authors: Ognjen Arandjelovic, Duc-Son Pham, Svetha Venkatesh

IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section we describe our evaluation of the proposed method, and report and discuss its performance in the context of the current state-of-the-art. We begin by describing our data set and evaluation protocol, follow with a presentation of a comprehensive set of performance statistics, and finish off with an analysis of our results and their significance.
Researcher Affiliation Academia Centre for Pattern Recognition & Data Analytics Department of Computing Deakin University, Australia Curtin University, Australia
Pseudocode No The paper describes the proposed approach and its steps in detail using textual explanations and mathematical equations, but it does not include a formal pseudocode block or algorithm box.
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository for the described methodology.
Open Datasets No It is an unsurprising consequence of this that there are no public data sets suitable for the evaluation of set-based algorithms so we collected a novel data set ourselves. Our data set comprises 10 image sets, each set containing 10 images acquired at different, non-uniformly distributed dates, as illustrated in Fig 5. This data was manually downloaded using the freely accessible web portal provided by Nearmap Ltd.
Dataset Splits No The paper mentions collecting a novel dataset and using 10 image sets, each containing 10 images, and also evaluates
Hardware Specification Yes The estimates are averages of 100 executions ran in Matlab 7 on an AMD Phenom II X4 965 processor with 8GB RAM.
Software Dependencies Yes ARRSI and the SURF-based methods were implemented primarily in C, with a Matlab wrapper , while the proposed method was implemented fully in Matlab. The estimates are averages of 100 executions ran in Matlab 7 on an AMD Phenom II X4 965 processor with 8GB RAM.
Experiment Setup Yes The particular filters we used in our experiments have the values of 40, 20, 8, and 3 pixels for the parameter σ in (7).