Object detection

Introduce

Object detection is used to identify the category and location of objects in images. It is a common problem encountered in the field of image vision. Regarding the issue of object detection. We provide an automated modeling solution. Enable users to solve object detection problems on their own.

Evaluating indicator

MAP - mean Average Precision

mAP Representing various categories AP The average value, and AP yes PR Area under the curve, about AP Explanation and specific calculation method reference: https://github.com/rafaelpadilla/Object-Detection-Metrics#average-precision

Parameter explanation

parameter	explain
image_height	(Required)Image height
image_width	(Required)image width
train_meta	(Required)Training set csv File Path
val_meta	(Required)Verification set csv File Path
test_meta	(Optional filling)Test set csv File Path
early_stopping_step	(Optional filling)Stop training steps in advance. The default is10

Practical examples

In this section, we will combine data formats. Provide effect explanations for several problem examples.

Defect detection

This question is about PCB Problem of finding defect points on the board. Locate the location of defects through object detection. Using automatic modeling to train and learn. In approximately100Within hours. The best model found. Its evaluation indicators mAP achieve 0.9756.

Wood testing

This question is about trucks in natural scenes. Conduct wood inspection. The purpose is to identify the number and location of wood on the truck. Training and learning through automatic modeling, mAP achieve 0.9452.