colab 数据集_Google Colab上的YOLOv4：轻松训练您的自定义数据集（交通标志）

colab 数据集

Traffic sign detection with YOLOv4

YOLOv4的交通标志检测

• Colab Notebook for training YOLOv4 with custom dataset (traffic signs)

  Colab Notebook，用于使用自定义数据集(交通标志)训练YOLOv4

大纲 (Outline)

• Introduction

  介绍

• Why YOLOv4?

  为什么选择YOLOv4？

• Data preparation

  资料准备

• Training with Colab

  用Colab进行培训

• Predict with YOLOv4

  用YOLOv4预测

• Conclusion

  结论

一，引言 (I. Introduction)

A while ago, I wrote a tutorial on training YOLOv3 with a custom dataset (gun detection) using the free GPU provided by Google Colab. After publishing the tutorial, many people emailed me asking about problems they faced during their training process. What I noticed is that the majority of those issues come from modifying inappropriate layer architecture, running cells in incorrect directory paths or missing one/some of the required configuration steps. I also realized that the setting up process might take time and often can be quite confusing.

前一段时间，我写了一个教程，介绍如何使用Google Colab提供的免费GPU使用自定义数据集(枪支检测)训练 YOLOv3。 发布教程后，许多人给我发了电子邮件，询问他们在培训过程中遇到的问题。 我注意到的是，大多数问题来自修改不合适的层体系结构，在错误的目录路径中运行单元或缺少一个/某些必需的配置步骤。 我还意识到设置过程可能会花费一些时间，并且常常会造成混乱。

For those reasons, I have came up with a new approach in which most of the setting up steps are automated with a python script (Colab allows you modifying and running python script directly on its environment). With this new procedure, the only thing you need to perform locally is to prepare your custom dataset following YOLO format, while all of the remains would be executed in Colab environment.

出于这些原因，我想出了一种新方法，其中大多数设置步骤都是使用python脚本自动化的(Colab允许您直接在其环境中修改和运行python脚本)。 使用此新过程，您只需在本地执行的操作就是按照YOLO格式准备自定义数据集，而所有其余数据都将在Colab环境中执行。

二。 为什么选择YOLOv4？ (II. Why YOLOv4?)