vision_processor.json

{
    "processor": {
        "name": "phi_4_vision_processor",
        "transforms": [
            {
                "operation": {
                    "name": "decode_image",
                    "type": "DecodeImage",
                    "attrs": {
                        "color_space": "RGB"
                    }
                }
            },
            {
                "operation": {
                    "name": "phi4_vision_dynamic_preprocess",
                    "type": "Phi4VisionDynamicPreprocess",
                    "attrs": {
                        "dynamic_hd": 36,
                        "dyhd_base_resolution": 448
                    }
                }
            },
            {
                "operation": {
                    "name": "rescale",
                    "type": "Rescale",
                    "inputs": [
                        ":0"
                    ]
                }
            },
            {
                "operation": {
                    "name": "normalize",
                    "type": "Normalize",
                    "attrs": {
                        "mean": [
                            0.5,
                            0.5,
                            0.5
                        ],
                        "std": [
                            0.5,
                            0.5,
                            0.5
                        ]
                    }
                }
            },
            {
                "operation": {
                    "name": "phi4_vision_processor",
                    "type": "Phi4VisionProcessor",
                    "inputs": [
                        ":0",
                        "phi4_vision_dynamic_preprocess:1"
                    ],
                    "attrs": {
                        "dyhd_base_resolution": 448,
                        "interpolation": "CUBIC"
                    }
                }
            }
        ],
        "output_aligner": "phi4-vision-aligner"
    }
}