Darknet convolutional-padding allocation DoS PoC
This repository contains a benign security research PoC for a Darknet config
that drives large allocation in cv2.dnn.readNetFromDarknet(...).
Files:
control_same_size.cfgmalicious_padding_20000.cfgtiny_shared.weightsreproduce.py
Observed behavior:
- control config:
- loads successfully
- malicious config:
- with
[convolutional] padding=20000 - raises:
Failed to allocate 6400320004 bytes
- with
Reproduction:
python3 build_poc.py
python3 reproduce.py
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support