解决[DNNDK] invalid pixel value of input tensor:128

七七库咪呀

已于 2023-01-04 09:55:13 修改

阅读量200

点赞数

分类专栏： FPGA学习笔记文章标签：前端编辑器 html

于 2023-01-04 09:51:05 首次发布

本文链接：https://blog.csdn.net/qq_42870876/article/details/128542949

版权

FPGA学习笔记专栏收录该内容

4 篇文章 0 订阅

订阅专栏

解决[DNNDK] invalid pixel value of input tensor:128

vitis ai yolo的上板bug

解决方式：
将image_data /= 255.改成mage_data /= 512.

'''image preprocessing'''
def pre_process(image, model_image_size):
    image = image[...,::-1]
    image_h, image_w, _ = image.shape
 
    if model_image_size != (None, None):
        assert model_image_size[0]%32 == 0, 'Multiples of 32 required'
        assert model_image_size[1]%32 == 0, 'Multiples of 32 required'
        boxed_image = letterbox_image(image, tuple(reversed(model_image_size)))
    else:
        new_image_size = (image_w - (image_w % 32), image_h - (image_h % 32))
        boxed_image = letterbox_image(image, new_image_size)
    image_data = np.array(boxed_image, dtype='float32')
    image_data /= 255.///////////改这里////////////
    image_data = np.expand_dims(image_data, 0) 	
    return image_data