zoukankan      html  css  js  c++  java
  • 物体检测,Error: maximum box coordinate value is too large

    使用ssd目标检测,出现error:maximum box coordinate value is larger than 1.100000: ] [1.325]

    主要原因在于,用labelImg 标记的目标太小,以及标记工具出现问题

    1 deleted all boxes that are less than 1/16 th of the image size and the training works fine.,https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10/blob/master/sizeChecker.py

    if x_value < 33 or y_value < 33: 删除,shutil.move(file, tree + '/wrong_data/')

    2.忽视结果

    Actually, I just ignore this error, change the ./core/box_list_ops.py
    `max_assert = tf.Assert(tf.greater_equal(1.1, box_maximum), ['maximum box coordinate value is larger ' 'than 1.1: ', box_maximum])`
    1.01 to 1.1, as my error max value is 1.014.
    By this way to make the code keep on. Seems the result is not affected.

    参考链接:https://github.com/tensorflow/models/issues/1754

  • 相关阅读:
    Java/android下哈希sha1和MD5的实现
    ANDROID SOCKET 开发
    UML补充
    TCP协议中的三次握手和四次挥手(转)
    uva 658 最短路
    uva 11280 最短路
    uva 10246 最短路
    uva 11747,kruskal 并查集
    uva 544 dijkstra
    uva 1395 瓶颈树
  • 原文地址:https://www.cnblogs.com/walktosee/p/10371810.html
Copyright © 2011-2022 走看看