Enhancing Data Accessibility for Reproducibility in Animal Detection Experiments

[yihong1120]

Dear Authors,

I trust this message finds you in good health and spirits. I am writing to address a matter of significance regarding the reproducibility of experiments detailed in your repository, which accompanies the paper "Animal Detection in Man-made Environments".

Firstly, I would like to express my appreciation for the comprehensive nature of your repository. The inclusion of modified versions of open-source repositories and the hierarchical organisation of commands is commendable. It greatly facilitates the understanding and execution of the experiments.

However, I have encountered a challenge that I believe may be affecting others as well. The data necessary to replicate the experiments is hosted on OneDrive, which, while functional, presents certain limitations in terms of accessibility and download stability. This can be particularly problematic for researchers with limited internet connectivity or those attempting to automate the download process for large-scale replication studies.

To enhance the accessibility and ease of use for the research community, I propose the following suggestions:

1. Mirroring Data on Additional Platforms: Consider hosting the dataset on multiple platforms, such as Zenodo or Figshare, which offer robust support for academic data sharing and are often used for archiving datasets associated with research papers.

2. Data Versioning: Implement a version control system for the dataset to ensure that changes and updates to the data are tracked, allowing researchers to access specific versions referenced in experiments.

3. Checksums for Data Integrity: Provide checksums (e.g., MD5 or SHA-256) for the dataset files, enabling researchers to verify the integrity of the data post-download, ensuring that the files have not been corrupted or altered.

4. Command-Line Download Scripts: Offer a script to facilitate the command-line download of the dataset, which can be particularly useful for researchers working on remote servers or those who prefer automated data retrieval methods.

I believe that addressing these points would significantly bolster the reproducibility of your experiments and further the impact of your research. I am more than willing to assist in the implementation of these enhancements should you require it.

Thank you for considering my suggestions. I look forward to your response and am eager to engage further with your work.

Best regards,
yihong1120