From 40f7f1f9307d6196b7c12302a5e3d47076c2ce1c Mon Sep 17 00:00:00 2001 From: Kaipeng Deng Date: Wed, 28 Aug 2019 19:38:24 +0800 Subject: [PATCH] fix yolov3 readme (#3213) --- PaddleCV/yolov3/README.md | 2 +- PaddleCV/yolov3/README_en.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/PaddleCV/yolov3/README.md b/PaddleCV/yolov3/README.md index 689cdb5c..4f4925d2 100644 --- a/PaddleCV/yolov3/README.md +++ b/PaddleCV/yolov3/README.md @@ -220,7 +220,7 @@ YOLOv3检测原理 ### 模型结构 -YOLOv3将输入图像分成S\*S个格子，每个格子预测B个bounding box，每个bounding box预测内容包括: Location(x, y, w, h)、Confidence Score和C个类别的概率，因此YOLOv3输出层的channel数为S\*S\*B\*(5 + C)。YOLOv3的loss函数也有三部分组成：Location误差，Confidence误差和分类误差。 +YOLOv3将输入图像分成S\*S个格子，每个格子预测B个bounding box，每个bounding box预测内容包括: Location(x, y, w, h)、Confidence Score和C个类别的概率，因此YOLOv3输出层的channel数为B\*(5 + C)。YOLOv3的loss函数也有三部分组成：Location误差，Confidence误差和分类误差。 YOLOv3的网络结构如下图所示:

diff --git a/PaddleCV/yolov3/README_en.md b/PaddleCV/yolov3/README_en.md index 142903bf..cb6f2b5b 100644 --- a/PaddleCV/yolov3/README_en.md +++ b/PaddleCV/yolov3/README_en.md @@ -221,7 +221,7 @@ YOLOv3 detection principle ### Model structure -YOLOv3 divides the input image in to S\*S grids and predict B bounding boxes in each grid, predictions of boxes include Location(x, y, w, h), Confidence Score and probabilities of C classes, therefore YOLOv3 output layer has S\*S\*B\*(5 + C) channels. YOLOv3 loss consists of three parts: location loss, confidence loss and classification loss. +YOLOv3 divides the input image in to S\*S grids and predict B bounding boxes in each grid, predictions of boxes include Location(x, y, w, h), Confidence Score and probabilities of C classes, therefore YOLOv3 output layer has B\*(5 + C) channels. YOLOv3 loss consists of three parts: location loss, confidence loss and classification loss. The bone network of YOLOv3 is darknet53, the structure of YOLOv3 is as follow:


-- GitLab