像素级监督疑惑

[csjgit123]

深度图这一部分网络最后的输入是一副特征图？实际应用时怎么做分类呢？标量监督最后会给出0或1代表预测的物体类别，那么像素级监督最后的输出是特征图要怎么做分类呢？

[csjgit123]

if result_mean < 0.5:
        return 0
    else:
        return 1

为什么我认为<0.5应该是1呢，真实人脸使用的深度图监督，那么欺骗人脸使用全黑的图监督，欺骗人脸得到的特征图的均值应该会更小一些吧，那么result_mean<0.5应该认为是欺骗类别吧？

[csjgit123]

    test_dir = "/home/shicaiwei/data/liveness_data/CASIA-FASD/test/living"
    label = 0
    pre_path = "../output/models/depth_patch.pth"
    isface = True
    depth_cnn_test(pre_path=pre_path, test_dir=test_dir, label=label,
                   isface=isface)

depth_cnn_test.py中，这里的label是否应该是1？

[shicaiwei123]

是的，living face 对应是1，我会更正一下。谢谢。

电子科技大学 信息与通信工程院 School of Information and Communication Engineering University of Electronic Science and Technology of China (UESTC) +18285516726 韦仕才

发自 网易邮箱大师

---- Replied mail ---- | From | @.> | | Date | 12/05/2021 11:28 | | To | @.> | | Cc | @.***> | | Subject | Re: [shicaiwei123/patch_based_cnn] 像素级监督疑惑 (Issue #7) |

test_dir = "/home/shicaiwei/data/liveness_data/CASIA-FASD/test/living"

label = 0

pre_path = "../output/models/depth_patch.pth"

isface = True

depth_cnn_test(pre_path=pre_path, test_dir=test_dir, label=label,

               isface=isface)

depth_cnn_test.py中，这里的label是否应该是1？

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.