ashim/floodmask-crp-classifier-2026-03-06

FloodMask CRP Classifier (Credential-Requiring Page)

Part of the FloodMask phishing detection pipeline. Determines whether a webpage is actively requesting credentials -- the final gate before issuing a phishing verdict.

Pipeline Role

Screenshot -> [AWL] -> element regions -> [Logo Classifier] -> brand match -> [CRP Classifier] -> phishing verdict

A page that matches a known brand logo but does not request credentials is not classified as an active phishing page. The CRP classifier filters these cases out.

Model Details

• Format: ONNX (optimized for CUDA via optimum.onnxruntime)
• Input: Webpage screenshot, resized to 224x224, normalized via ToTensor()
• Output: Binary -- credential-requiring (1) or not (0)

Preprocessing

This model uses a torchvision.transforms pipeline rather than a HuggingFace feature extractor:

from torchvision import transforms

processor = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
])

from PIL import Image
image = Image.open("screenshot.png").convert("RGB")
tensor = processor(image).unsqueeze(0)  # add batch dim

Usage

from optimum.onnxruntime import ORTModel

model = ORTModel.from_pretrained("ashim/floodmask-crp-classifier-2026-03-06")
outputs = model(pixel_values=tensor)

CUDA Usage

CUDA_OPTIONS = {
    "device_id": 0,
    "arena_extend_strategy": "kNextPowerOfTwo",
    "gpu_mem_limit": 2 * 1024 * 1024 * 1024,
    "cudnn_conv_algo_search": "EXHAUSTIVE",
    "do_copy_in_default_stream": True,
}

model = ORTModel.from_pretrained(
    "ashim/floodmask-crp-classifier-2026-03-06",
    provider="CUDAExecutionProvider",
    provider_options=CUDA_OPTIONS,
)

Related Models

• ashim/floodmask-awl-2026-03-06
• ashim/floodmask-logo-classifier-2026-03-06

Citation

@inproceedings{liu2022phishintention,
  title={PhishIntention: Toward Explainable Content-based Phishing Detection through Intent Identification},
  author={Liu, Ruofan and Lin, Yun and Yang, Xianglin and Ng, Siang Hwee and Divakaran, Dinil Mon and Dong, Jin Song},
  booktitle={USENIX Security Symposium},
  year={2022}
}