[英]CNN Object detection: How to reduce high false positive rates
I am currently using Faster RCNN with inception v2 on Tensorflow object detection API on WIDER-FACE data, i have a lot of false positives with high scores (>0.98, so setting a higher score threshold won't help).我目前在 WIDER-FACE 数据上的 Tensorflow 对象检测 API 上使用 Faster RCNN 和 inception v2,我有很多高分的误报(> 0.98,因此设置更高的分数阈值无济于事)。 I have already assigned Hard Example Mining in my code but it doesn't help to much.
我已经在我的代码中分配了困难示例挖掘,但它没有多大帮助。 For image preprocessing, I randomly crop an area from the original image using tf.image.sample_distorted_bounding_box and resize it to 300*300.
对于图像预处理,我使用tf.image.sample_distorted_bounding_box从原始图像中随机裁剪了一个区域,并将其大小调整为 300*300。 The resized image will be randomly flipped by a probability of 0.5
调整大小的图像将以 0.5 的概率随机翻转
I set batch size to 32 and for each image the ratio of positive:negative in hard negative mining operation is 32:32 .我将批量大小设置为 32 并且对于每个图像,硬负挖掘操作中的正:负比为32:32 。 I set IoU > 0.5 as positive and IoU < 0.3 as negative.
我设置IoU > 0.5为正, IoU < 0.3为负。 For samples that IoU between 0.5 and 0.3 are ignored
对于 IoU 介于 0.5 和 0.3 之间的样本被忽略
Can anyone help me with this?谁能帮我这个? Many thanks!
非常感谢!
Usually the ratio between positive and negative examples is set to be 1:3.通常正例和反例的比例设置为 1:3。 It could be that because in your model the ratio is set to 1:1, the model doesn't see enough negative examples.
可能是因为在您的模型中比率设置为 1:1,模型没有看到足够的反例。
You can also try to do an error analysis and check what are the false positives which have high confidence.您还可以尝试进行错误分析并检查哪些是具有高置信度的误报。 Maybe they are of a specific type, and if so - add negative examples of this type to your training data.
也许它们是特定类型的,如果是这样 - 将这种类型的反例添加到您的训练数据中。
您也可以尝试增加 RPN 的 NMS 阈值。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.