Executive brief
The imgaug library, a tool used to create large datasets for machine learning by altering images, contains a security flaw in its multicore processing component. An attacker could exploit this to execute malicious code on the system running the library by providing specially crafted data. This could lead to a complete system takeover or unauthorized access to sensitive machine learning data and infrastructure.
Technical details
An insecure deserialization vulnerability exists in the imgaug library (up to version 0.4.0) within the BackgroundAugmenter class of the multicore.py module. The _augment_images_worker() method utilizes Python's pickle module to deserialize data from a multiprocessing queue without validation. An attacker who can control or influence the data placed into this queue—via malicious input scripts, social engineering, or a compromised shared queue—can inject a malicious pickle payload. Upon deserialization, this payload executes arbitrary code in the context of the worker process. As of the advisory date, no patched versions are available.
Affected products
- aleju imgaug <= 0.4.0
Timeline
- 2026-05-12: advisory: GitHub Advisory published
- 2026-05-12: disclosed: NVD published date