hb_verifier -m quantized_nv12_remove_stage3.bc,quantized_nv12_remove_stage3.hbm -i y_data.npy,uv_data.npy

hb_model_info quantized_nv12_remove_stage3.bchb_model_info quantized_nv12_remove_stage3.hbm

from PIL import Imageimport numpy as np
def generate_nv12(img):    w,h = img.size    # Convert images to YUV format    yuv_img = img.convert('YCbCr')    y_data, u_data, v_data = yuv_img.split()
    # Convert Y, U, and V channel data to byte streams    y_data_bytes = y_data.tobytes()    u_data_bytes = u_data.resize((u_data.width // 2, u_data.height // 2)).tobytes()    v_data_bytes = v_data.resize((v_data.width // 2, v_data.height // 2)).tobytes()
    # Arrange the UV data in the form of UVUVUVUV...     uvuvuv_data = bytearray()    for u_byte, v_byte in zip(u_data_bytes, v_data_bytes):        uvuvuv_data.extend([u_byte, v_byte])
    # Input for the hbir model    y = np.frombuffer(y_data_bytes, dtype=np.uint8).reshape(1, h, w, 1).astype(np.uint8)    np.save("y_data.npy", y)    uv = np.frombuffer(uvuvuv_data, dtype=np.uint8).reshape(1, h//2, w//2, 2).astype(np.uint8)    np.save("uv_data.npy", uv)    return y, uv
if __name__ == "__main__":    # Create a random image with the shape (1, 512, 960, 3)    # Generate random RGB values in the range 0-255    image_data = np.random.randint(0, 256, (512, 960, 3), dtype=np.uint8)
    # Convert the numpy array to a PIL image    img = Image.fromarray(image_data)    y, uv = generate_nv12(img)

import numpy as np# 定义所有输入数据的形状和数据类型data_dict = {    'A_input': {'shape': (1, 32, 16), 'dtype': np.int8},    'B_input': {'shape': (1, 1, 320, 4), 'dtype': np.int16},    'C_input': {'shape': (32, 10, 10), 'dtype': np.bool_},}
# 生成数据并保存为 .npy 文件for idx, (name, info) in enumerate(data_dict.items()):    shape = info['shape']    dtype = info['dtype']    # 生成随机数据    data = np.random.randint(0, 100, size=shape, dtype=dtype) if dtype == np.int16 else np.random.choice([True, False], size=shape)    # 保存为 .npy 文件，文件名前加上编号    np.save(f"{idx}_{name}.npy", data)    print(f"Saved {idx}_{name}.npy")

hb_verifier -m quantized_nv12_remove_stage3.bc,quantized_nv12_remove_stage3.hbm -i y_data.npy,uv_data.npy

from hbdk4.compiler import load, Hbmimport numpy as npfrom PIL import Image
def generate_nv12(img):    w,h = img.size    # Convert images to YUV format    yuv_img = img.convert('YCbCr')    y_data, u_data, v_data = yuv_img.split()
    # Convert Y, U, and V channel data to byte streams    y_data_bytes = y_data.tobytes()    u_data_bytes = u_data.resize((u_data.width // 2, u_data.height // 2)).tobytes()    v_data_bytes = v_data.resize((v_data.width // 2, v_data.height // 2)).tobytes()
    # Arrange the UV data in the form of UVUVUVUV...     uvuvuv_data = bytearray()    for u_byte, v_byte in zip(u_data_bytes, v_data_bytes):        uvuvuv_data.extend([u_byte, v_byte])
    # Input for the hbir model    y = np.frombuffer(y_data_bytes, dtype=np.uint8).reshape(1, h, w, 1).astype(np.uint8)    # np.save("y_data.npy", y)    uv = np.frombuffer(uvuvuv_data, dtype=np.uint8).reshape(1, h//2, w//2, 2).astype(np.uint8)    # np.save("uv_data.npy", uv)    return y, uv
def compare_arrays(array1, array2, decimal_places=2):    """    Compare two arrays for consistency up to a specified number of decimal places.
    Parameters:    - array1: First numpy array.    - array2: Second numpy array.    - decimal_places: Number of decimal places to consider for alignment.
    Returns:    - are_equal: True if arrays are consistent up to the specified decimal places, False otherwise.    - max_difference: Maximum difference (absolute value) if arrays are not consistent, else 0.    """    # Round the arrays to the specified decimal places    rounded1 = np.round(array1, decimals=decimal_places)    rounded2 = np.round(array2, decimals=decimal_places)        # Check equality    are_equal = np.array_equal(rounded1, rounded2)        # Calculate maximum difference if not equal    max_difference = 0    if not are_equal:        max_difference = np.max(np.abs(array1 - array2))        return are_equal, max_difference
hbir = load("./quantized_nv12_remove_stage3.bc")hbm = Hbm("./quantized_nv12_remove_stage3.hbm")
# Create a random image with the shape (1, 512, 960, 3)# Generate random RGB values in the range 0-255image_data = np.random.randint(0, 256, (512, 960, 3), dtype=np.uint8)# Convert the numpy array to a PIL imageimg = Image.fromarray(image_data)y, uv = generate_nv12(img)
inputs = {"input_0_y": y, "input_0_uv": uv}
# 分别进行hbir和Hbm推理hbir_outputs = hbir[0].feed(inputs)# print("hbir_outputs:", hbir_outputs)hbm_x86_outputs = hbm[0].feed(inputs)# print("hbm_x86_outputs:", hbm_x86_outputs)
# 比较Hbir和hbm输出for idx, v in enumerate(hbir[0].outputs):    hbir_data = hbir_outputs[v.name]    hbm_x86_data = hbm_x86_outputs[v.name]
    # Compare arrays    are_equal, max_difference = compare_arrays(hbir_data, hbm_arrch64_data, decimal_places=4)    if not are_equal:        print("Maximum difference:", max_difference)    else:        print(f"outputs[{idx}] is equal!")