减少预训练 Tensorflow 模型检测到的对象数量 [英] Reduce number of objects a pretrained Tensorflow model detects
问题描述
我正在使用此代码作为对象即使在大多数图片中有 0-5 个对象,它也会输出 100 个框.在 250X250 图像上检测需要 5 秒.减少检测对象的数量是否会加快这个过程,如果是,有没有办法做到这一点?
I am using this code for object detection and it outputs 100 boxes even though in most pictures there are 0-5 objects. The detection takes 5 seconds on a 250X250 image. Would cutting the number of objects to be detected speed up the process, and if yes is there a way to do it?
推荐答案
从逻辑上讲是可以的,我不能说具体是多少.您可以将模型重新训练到您感兴趣的较少对象.对于训练,您还将指定对扩展名为 *.pbtxt 的文件感兴趣的对象,该文件包含指定的对象列表.
Logically it would, i cant say exactly by how much. You can retrain the model to only fewer objects that you are interested in. For training you will also specify objects interested in a file with extension *.pbtxt which has list of objects specified.
很少有详细讨论再培训的链接https://pythonprogramming.net/training-custom-objects-tensorflow-object-detection-api-tutorial/
Few Links which talk about retraining in detail https://pythonprogramming.net/training-custom-objects-tensorflow-object-detection-api-tutorial/
https://medium.com/@dana.yu/training-a-custom-object-detection-model-41093ddc5797
这篇关于减少预训练 Tensorflow 模型检测到的对象数量的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!