Protecting Geolocation Privacy of Photo Collections
Authors: Jinghan Yang, Ayan Chakrabarti, Yevgeniy Vorobeychik524-531
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Through extensive experiments on real photo collections, we demonstrate that our approaches are indeed highly effective at preserving geolocation privacy. |
| Researcher Affiliation | Academia | Jinghan Yang, Ayan Chakrabarti, Yevgeniy Vorobeychik Computer Science & Engineering, Washington University in St. Louis {jinghan.yang, ayan, yvorobeychik}@wustl.edu |
| Pseudocode | No | The paper describes algorithms and methods in textual form but does not provide any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Our code and extended version could be found in the project page 2. https://github.com/jinghan Y/geo Privacy Album |
| Open Datasets | No | We conduct experiments on a set of 750k images downloaded from Flickr, where we ensure each image has a geotag to specify groundtruth location. Further, we collected 10k Flickr user designated albums. The paper mentions using images downloaded from Flickr and user-designated albums, but does not provide a direct link, DOI, or formal citation for this specific collected dataset to make it publicly accessible or reproducible. |
| Dataset Splits | No | We randomly split these into training and test sets, of size 90% and 10% of total images respectively. The paper only mentions training and test sets, with no explicit mention of a validation set split or how it was used for hyperparameter tuning. |
| Hardware Specification | No | The paper mentions support from NVIDIA, which implies GPU usage, and the use of the CPLEX solver, but it does not specify any particular GPU models, CPU models, or other detailed hardware specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions using a VGG-16 architecture and the CPLEX solver, but it does not specify version numbers for these or any other software dependencies, such as programming languages, deep learning frameworks, or libraries. |
| Experiment Setup | No | The paper describes the VGG-16 network architecture, loss function (cross-entropy), and mentions initializing weights from an ImageNet-trained model, but it lacks specific training hyperparameters such as learning rate, batch size, number of epochs, or optimizer details for their experiments. |