Overview
Who this competition fits
适合已有计算机视觉目标检测基础、愿意使用 PaddlePaddle 2.0 及以上版本训练端到端模型的个人参赛者。你需要能处理钢铁表面灰度图像,并输出缺陷类别、边界框坐标和置信度。
Read the original official blurb
本次比赛聚焦图像目标识别技术,需要选手从图像中识别出钢铁表面的缺陷位置,并给出锚点框的坐标,同时对不同的缺陷进行分类,以期产出泛化性更好、性能更稳定的钢铁表面缺陷识别模型。 Official tags: 图像目标识别. Sign-ups: 1396.
Resources
What this competition gives you
Taken from the competition's official page.
Preparation
From registration to a first submission
- 01
PaddlePaddle 2.0+
- 02
目标检测模型
- 03
边界框标注与坐标
- 04
图像缺陷分类
- 05
CSV结果文件生成
- 06
mAP与IoU评估
Before you commit: 难点在于同时做好缺陷定位和六类缺陷分类:同类缺陷的形态、方向和灰度会变化,而不同类别之间也需要区分。提交还必须生成严格格式的 submission.csv,且每个检测框都要单独成行。
Source
How this page was assembled
Competition information is structured from the official page. Scores are platform estimates for decision support; official rules take precedence.
- Official competition page
- Baidu AI Studio
- Last checked
- Not recorded