Solution: Object Detection
Introduction
Object detection is used to identify the categories and locations of objects in images, which is a frequently encountered problem in the field of computer vision. For object detection problems, we provide an automated modeling solution that enables users to solve object detection problems they encounter on their own.
Evaluation Metrics
mAP - mean Average Precision
mAP represents the mean of Average Precision across all categories, where AP is the area under the PR curve. For explanations and specific calculation methods of AP, please refer to: https://github.com/rafaelpadilla/Object-Detection-Metrics#average-precision
Parameter Explanation
| Parameter | Description |
|---|---|
| image_height | (Required) Image height |
| image_width | (Required) Image width |
| train_meta | (Required) Training set CSV file path |
| val_meta | (Required) Validation set CSV file path |
| test_meta | (Optional) Test set CSV file path |
| early_stopping_step | (Optional) Early stopping training steps, default is 10 |
Practical Examples
In this section, we will combine data formats with several problem examples to demonstrate the results.
Defect Detection
This problem involves finding defect points on PCB boards by using object detection to locate the positions of defects. Using automated modeling for training and learning, within approximately 100 hours, the optimal model found achieved an evaluation metric mAP of 0.9756.
Timber Detection
This problem involves detecting timber on trucks in natural scenes, with the goal of identifying the number and positions of timber pieces on the truck. After automated modeling training and learning, the mAP reached 0.9452.