eye-privacy
Science Score: 44.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
✓CITATION.cff file
Found CITATION.cff file -
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (5.1%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: ddkul
- License: mit
- Language: Python
- Default Branch: main
- Size: 40 KB
Statistics
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Eye-Privacy
This repository emulates a privacy-preserving framework for transferring images of an individual's eye to a remote model for gaze and emotion detection. This method combines the principles of the additive secret-sharing [1] and Bluetooth transmission schemes [2]. In this strategy, the image to be transmitted is split into a fixed number of pieces, and these splits are sent to a random subset of intermediate servers, amongst a set of available servers. These intermediate servers then forward the received splits to the remote model server.
We use Docker Engine to simulate seven containers, each behaving as an independent server, connected to the same network. Out of the six servers, one was the application server, one was the remote model server, one was an adversarial server, while the remaining four functioned as the intermediate transmitting servers A, B, C, and D. The servers use APIs to communicate with each other built using FastAPI. To demonstrate the privacy scheme, we set up a replay attack scenario. The application server split and sent two images, one after another, according to the identification number and servers' sequence. A replay attacker captures the split that arrives at server A, and replays it when the second image is sent. It is assumed that an out-of-band channel is used to exchange the random number generator seed for identification numbers and corresponding sequences between the application server and remote model server.
To simulate -
python
docker-compose build
docker-compose up
To retrieve images in the model server container (update path) -
python
sh copy_images.sh
The endpoint to split images is available on http://0.0.0.0/splitimgs
The endpoint for replaying the first image's split once it has been received at the model server is http://0.0.0.0:90/replaysplit.
References
[1] Adi Shamir. “How to Share a Secret”. In: Commun. ACM 22.11 (1979), 612–613.
[2] Jaap Haartsen. “Bluetooth-The universal radio interface for ad hoc, wireless connectivity”. In: Ericsson review 3.1 (1998), pp. 110–117.
Owner
- Name: Divya Kulkarni
- Login: ddkul
- Kind: user
- Repositories: 1
- Profile: https://github.com/ddkul
Citation (CITATION.cff)
# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!
cff-version: 1.2.0
title: >-
A method for preserving the privacy of eye images
used for emotion detection.
message: >-
If you use this software, please cite it using the
metadata from this file.
type: software
authors:
- given-names: Divya
name-particle: D
family-names: Kulkarni
- given-names: Sougata
family-names: Sen