35 lines
1.0 KiB
Python
35 lines
1.0 KiB
Python
|
from resnet import build_resnet
|
||
|
|
import cv2
|
||
|
|
import torchvision.transforms as transforms
|
||
|
|
|
||
|
|
|
||
|
|
backbone, _ = build_resnet(name="resnet50", pretrained=True)
|
||
|
|
|
||
|
|
# 读取测试图像
|
||
|
|
image = cv2.imread('E:/DeepLearning/PersonSearch/COAT/COAT/main/backbone/test.jpg') # 替换为您的测试图像文件路径
|
||
|
|
image = cv2.resize(image, (224, 224))
|
||
|
|
|
||
|
|
# 使用 OpenCV 显示原始图像
|
||
|
|
cv2.imshow("Original Image", image)
|
||
|
|
cv2.waitKey(0)
|
||
|
|
cv2.destroyAllWindows()
|
||
|
|
# 创建转换以将图像缩放到 224x224 大小并转换为张量
|
||
|
|
transform = transforms.Compose([
|
||
|
|
transforms.ToPILImage(),
|
||
|
|
transforms.Resize((224, 224)),
|
||
|
|
transforms.ToTensor(),
|
||
|
|
])
|
||
|
|
|
||
|
|
# 应用转换并将图像转换为张量
|
||
|
|
image_tensor = transform(image)
|
||
|
|
image_tensor = image_tensor.unsqueeze(0) # 添加批次维度，将形状变为 [1, 3, 224, 224]
|
||
|
|
|
||
|
|
# 使用模型进行推理
|
||
|
|
output = backbone(image_tensor)
|
||
|
|
|
||
|
|
# 输入的shape是[1,1024,7,7]
|
||
|
|
#print(output)
|
||
|
|
print(output['feat_res4'].shape)
|
||
|
|
# 在这里，您可以处理模型的输出，进行后续的操作或分析
|
||
|
|
|