open-mmlab / OpenMMLabCamp

Apache License 2.0
93 stars 32 forks source link

【AI实战营第二期】第四次作业提交10班 #390

Open VoyagerXvoyagerx opened 1 year ago

VoyagerXvoyagerx commented 1 year ago

TO DO LIST:

  1. Labelme 标注语义分割数据集(子豪兄已经帮你完成了)
  2. 划分训练集和测试集(子豪兄已经帮你完成了)
  3. Labelme 标注转 Mask 灰度图格式(子豪兄已经帮你完成了)
  4. 使用 MMSegmentation 算法库,撰写 config 配置文件,训练 PSPNet 语义分割算法
  5. 提交测试集评估指标
  6. 自己拍摄西瓜图片和视频,将预测结果发到群里
  7. (选做)训练 Segformer 语义分割算法,提交测试集评估指标

西瓜瓤、西瓜籽数据集:

标注:同济子豪兄
image

image

类别名称 | 类别语义 | 标注类别 | 灰度图像素值 -- | -- | -- | -- / | 背景 | / | 0 red | 西瓜红瓤 | 多段线(polygon) | 1 green | 西瓜外壳 | 多段线(polygon) | 2 white | 西瓜白皮 | 多段线(polygon) | 3 seed-black | 西瓜黑籽 | 多段线(polygon) | 4 seed-white | 西瓜白籽 | 多段线(polygon) | 5

数据集下载链接:

需提交的测试集评估指标:(不能低于 baseline 指标的 50% 

aAcc: 60.6200
mIoU: 21.1400
mAcc: 28.4600

mm-assistant[bot] commented 1 year ago

We recommend using English or English & Chinese for issues so that we could have broader discussion.

AnakinYang commented 1 year ago

MMSegmentation 作业 WeChat ID:Anakin 作业链接:https://github.com/AnakinYang/homework/blob/main/mmseg.ipynb 国内Github可能较慢,备用链接:https://featurize.cn/notebooks/0acb51ec-cee9-4dcb-a28d-b9124fb75a69

aAcc: 89.8600 mIoU: 55.9000 mAcc: 61.9600 data_time: 0.0103 time: 0.2338 图片预测结果及评估指标见 notebook,视频预测见 Release,日志及相关配置文件见仓库。

Lin-zd commented 1 year ago

MMSeg:https://github.com/Lin-zd/Study/tree/main/MMSegmentation Wechat ID:柠檬酸 aAcc: 88.1200 mIoU: 53.0600 mAcc: 58.2200 data_time: 0.0135 time: 0.6841

chen-del commented 1 year ago

MMSeg:https://github.com/chen-del/mmlab-Camp-operation/tree/main/mmlab-camp-four Wechat ID: lo-yi aAcc: 92.8000 mIoU: 66.3700 mAcc: 76.8000 data_time: 0.1177 time: 1.1182 预测结果为:pred.jpg