DEF-rgbtcc: RGB-T Crowd Counting

Dual-Modulation Framework for RGB-T Crowd Counting via Spatially Modulated Attention and Adaptive Fusion.

Paper: ArXiv 2509.17079

Architecture

• Backbone: Shared VGG-19
• Encoder: Spatially Modulated Attention (SMA) Transformer
• Fusion: Adaptive Cross-Modal Fusion (ACMF)
• Output: Density map regression

Available Formats

• model.pth — PyTorch state dict
• model.safetensors — SafeTensors format
• model.onnx — ONNX (opset 17)
• model_fp16.trt — TensorRT FP16
• model_fp32.trt — TensorRT FP32

Usage

from def_rgbtcc.serve import RGBTCCInference

model = RGBTCCInference("model.pth")
result = model.predict(rgb_image, thermal_image)
print(f"Count: {result['count']:.1f}")

ANIMA Module

Part of the ANIMA Defense Module ecosystem (Wave 8). Products: ORACLE, ATLAS, NEMESIS