xView2-Solution
xView2损害评估挑战赛的第三名解决方案
尤金·赫维琴雅（Eugene Khvedchenya），2020年2月
该存储库包含用于解决源代码。 我的解决方案在公共LB上的得分为第二（0.803），在私人保留数据集上的得分为第三（0.805）。
简而言之
语义分割模型的集合。
经过加权CE培训，以解决班级不平衡问题。
进行大量增强，以防止过度拟合并增强对未对齐的前映像和后映像的鲁棒性。
前后图像共享编码器。 提取的特征将被串联并发送到解码器。
一堆编码器（ResNet，Densenet，EfficientNets）和两个解码器：Unet和FPN。
1轮伪标签
使用加权平均进行合奏。 在相应的验证数据上为每个模型优化的权重。
训练
从requirements.txt安装依赖项
关注train.sh
推理
要使用预训练的模型进行推断，请从“发布”选项卡下载完整档案，然后运行predi
文件下载
资源详情
[{"title":"（ 70 个子文件 3.79MB ） xView2-Solution","children":[{"title":"xView2-Solution-master","children":[{"title":"run_tensorboard_3389.sh <span style='color:#111;'> 101B </span>","children":null,"spread":false},{"title":"predict_oof.py <span style='color:#111;'> 3.84KB </span>","children":null,"spread":false},{"title":"evaluate_postprocessing.py <span style='color:#111;'> 3.54KB </span>","children":null,"spread":false},{"title":"test_docker_pytorch14_37.cmd <span style='color:#111;'> 1.63KB </span>","children":null,"spread":false},{"title":"train_folds.csv <span style='color:#111;'> 3.51MB </span>","children":null,"spread":false},{"title":"fit_predict.py <span style='color:#111;'> 24.22KB </span>","children":null,"spread":false},{"title":"convert_masks.py <span style='color:#111;'> 2.66KB </span>","children":null,"spread":false},{"title":"predict_37_weighted.py <span style='color:#111;'> 9.01KB </span>","children":null,"spread":false},{"title":"optimize_softmax.py <span style='color:#111;'> 2.82KB </span>","children":null,"spread":false},{"title":"train.sh <span style='color:#111;'> 11.49KB </span>","children":null,"spread":false},{"title":"predict.py <span style='color:#111;'> 4.54KB </span>","children":null,"spread":false},{"title":"finetune.py <span style='color:#111;'> 20.28KB </span>","children":null,"spread":false},{"title":"tests","children":[{"title":"test_damage_00121_prediction.png <span style='color:#111;'> 9.96KB </span>","children":null,"spread":false},{"title":"test_localization_00121_prediction.png <span style='color:#111;'> 12.29KB </span>","children":null,"spread":false},{"title":"test_post_00121.png <span style='color:#111;'> 1.55MB </span>","children":null,"spread":false},{"title":"pre.png <span style='color:#111;'> 1.97KB </span>","children":null,"spread":false},{"title":"test_postprocessing.py <span style='color:#111;'> 1.49KB </span>","children":null,"spread":false},{"title":"hurricane-florence_00000475_post_disaster.png <span style='color:#111;'> 7.70KB </span>","children":null,"spread":false},{"title":"test_dataset.py <span style='color:#111;'> 3.14KB </span>","children":null,"spread":false},{"title":"guatemala-volcano_00000000_post_disaster.png <span style='color:#111;'> 1.52MB </span>","children":null,"spread":false},{"title":"hurricane-florence_00000115_post_disaster.png <span style='color:#111;'> 9.72KB </span>","children":null,"spread":false},{"title":"post.png <span style='color:#111;'> 2.04KB </span>","children":null,"spread":false},{"title":"test_load_mask.py <span style='color:#111;'> 1.22KB </span>","children":null,"spread":false},{"title":"test_registration.py <span style='color:#111;'> 2.58KB </span>","children":null,"spread":false},{"title":"test_models.py <span style='color:#111;'> 3.70KB </span>","children":null,"spread":false}],"spread":false},{"title":"docker_submission_37.py <span style='color:#111;'> 9.91KB </span>","children":null,"spread":false},{"title":"convert_crops.py <span style='color:#111;'> 3.95KB </span>","children":null,"spread":false},{"title":"LICENSE <span style='color:#111;'> 1.03KB </span>","children":null,"spread":false},{"title":"black.toml <span style='color:#111;'> 524B </span>","children":null,"spread":false},{"title":"run_tensorboard.sh <span style='color:#111;'> 101B </span>","children":null,"spread":false},{"title":"build_push_docker_37.cmd <span style='color:#111;'> 244B </span>","children":null,"spread":false},{"title":"train.csv <span style='color:#111;'> 3.57MB </span>","children":null,"spread":false},{"title":"requirements.txt <span style='color:#111;'> 179B </span>","children":null,"spread":false},{"title":".gitignore <span style='color:#111;'> 1.22KB </span>","children":null,"spread":false},{"title":"make_folds.py <span style='color:#111;'> 1.66KB </span>","children":null,"spread":false},{"title":"requirements_docker_pytorch14.txt <span style='color:#111;'> 387B </span>","children":null,"spread":false},{"title":".dockerignore <span style='color:#111;'> 106B </span>","children":null,"spread":false},{"title":"README.md <span style='color:#111;'> 1.06KB </span>","children":null,"spread":false},{"title":"xview","children":[{"title":"models","children":[{"title":"common.py <span style='color:#111;'> 824B </span>","children":null,"spread":false},{"title":"fpn_v2.py <span style='color:#111;'> 7.26KB </span>","children":null,"spread":false},{"title":"unetv2.py <span style='color:#111;'> 10.45KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 1.28KB </span>","children":null,"spread":false},{"title":"cls.py <span style='color:#111;'> 1.79KB </span>","children":null,"spread":false}],"spread":false},{"title":"pseudo.py <span style='color:#111;'> 3.18KB </span>","children":null,"spread":false},{"title":"alignment.py <span style='color:#111;'> 2.82KB </span>","children":null,"spread":false},{"title":"ssim_loss.py <span style='color:#111;'> 2.05KB </span>","children":null,"spread":false},{"title":"train_utils.py <span style='color:#111;'> 2.38KB </span>","children":null,"spread":false},{"title":"optim.py <span style='color:#111;'> 6.93KB </span>","children":null,"spread":false},{"title":"augmentations.py <span style='color:#111;'> 6.68KB </span>","children":null,"spread":false},{"title":"inference.py <span style='color:#111;'> 17.43KB </span>","children":null,"spread":false},{"title":"metric.py <span style='color:#111;'> 8.46KB </span>","children":null,"spread":false},{"title":"dataset.py <span style='color:#111;'> 21.00KB </span>","children":null,"spread":false},{"title":"model_wrapper.py <span style='color:#111;'> 3.44KB </span>","children":null,"spread":false},{"title":"losses.py <span style='color:#111;'> 8.23KB </span>","children":null,"spread":false},{"title":"xview2_metrics.py <span style='color:#111;'> 11.09KB </span>","children":null,"spread":false},{"title":"factory.py <span style='color:#111;'> 4.61KB </span>","children":null,"spread":false},{"title":"postprocessing.py <span style='color:#111;'> 8.49KB </span>","children":null,"spread":false},{"title":"scheduler.py <span style='color:#111;'> 1.67KB </span>","children":null,"spread":false},{"title":"visualization.py <span style='color:#111;'> 4.71KB </span>","children":null,"spread":false},{"title":"rounder.py <span style='color:#111;'> 2.88KB </span>","children":null,"spread":false},{"title":"utils","children":[{"title":"combine_jsons.py <span style='color:#111;'> 4.84KB </span>","children":null,"spread":false},{"title":"inference.sh <span style='color:#111;'> 7.20KB </span>","children":null,"spread":false},{"title":"data_finalize.sh <span style='color:#111;'> 4.67KB </span>","children":null,"spread":false},{"title":"split_into_disasters.py <span style='color:#111;'> 4.88KB </span>","children":null,"spread":false},{"title":"view_polygons.ipynb <span style='color:#111;'> 12.37KB </span>","children":null,"spread":false},{"title":"inference_image_output.py <span style='color:#111;'> 7.73KB </span>","children":null,"spread":false},{"title":"mask_polygons.py <span style='color:#111;'> 11.19KB </span>","children":null,"spread":false}],"spread":false},{"title":"averaging_rounder.py <span style='color:#111;'> 3.19KB </span>","children":null,"spread":false}],"spread":false},{"title":"Dockerfile-pytorch14-37 <span style='color:#111;'> 572B </span>","children":null,"spread":false},{"title":"run_tensorboard.cmd <span style='color:#111;'> 121B </span>","children":null,"spread":false}],"spread":false}],"spread":true}]
评论信息
其他资源
- 0-9十个数字的中文和英文语音文件wav格式
- MATLAB 教学视频：详解快速傅里叶变换FFT在MATLAB中的实现
- DisplayPort1.2官方标准
- 北京大学数字逻辑设计实验课程讲义（2018年）
- 药品管理系统（c#）
- 4G LTE NAS信令3gpp标准技术文档（24301-g10中英对照.docx）
- Terraria源码英文-支持中文
- CC++学习指南 - 应用篇（线程与网络)
- 教育技术学核心期刊
- dobble_card_generator：生日快乐，乔纳斯:)-源码
- 02_HelloQt.rar
- 1_FSPT字典转换.txt
- Google play billing(Google play 内支付) 代码封装 api V1.1.3
- CNCAP-2012(中文版)
- javaweb新闻管理系统
- QTP 运行报:Microslft Visual C++ Runtime Library Runtime Error R6025解决方案
- IoT 2020: 智能安全的物联网平台 英文原版
- 项目源码（java项目TT商城）
- 一个ajax做的联想查询文本框(脚本)
- Android蓝牙连接设备demo
- RFID RSSI 论文
- 背包问题之贪婪算法求解C语言源代码).
- FIFO代码以及对应的testbench
- Visio模版
- 传智播客贴吧源码
免责申明
【只为小站】的资源来自网友分享，仅供学习研究，请务必在下载后24小时内给予删除，不得用于其他任何用途，否则后果自负。基于互联网的特殊性，【只为小站】 无法对用户传输的作品、信息、内容的权属或合法性、合规性、真实性、科学性、完整权、有效性等进行实质审查；无论 【只为小站】 经营者是否已进行审查，用户均应自行承担因其传输的作品、信息、内容而可能或已经产生的侵权或权属纠纷等法律责任。
本站所有资源不代表本站的观点或立场，基于网友分享，根据中国法律《信息网络传播权保护条例》第二十二条之规定，若资源存在侵权或相关问题请联系本站客服人员，zhiweidada#qq.com，请把#换成@，本站将给予最大的支持与配合，做到及时反馈和处理。关于更多版权及免责申明参见 版权及免责申明
本站所有资源不代表本站的观点或立场，基于网友分享，根据中国法律《信息网络传播权保护条例》第二十二条之规定，若资源存在侵权或相关问题请联系本站客服人员，zhiweidada#qq.com，请把#换成@，本站将给予最大的支持与配合，做到及时反馈和处理。关于更多版权及免责申明参见 版权及免责申明