baidu-remote-sensing

百度遥感比赛，前百分之一的实验方案

multimodal_train_val.py multimodal_train_val1.py multimodal_train_val2.py multimodal_train_val3.py分别对多个模型进行训练，其中有使用单GPU和多GPU

MM_val.py MM_val1.py MM_val2.py 分别对模型进行单模的验证，并用confusion matrix可视化数据，观察每个模型数据的学习情况，对模型进行筛选，为下一步集成做准备

GeResult_ensemble.py 对模型进行集成，最多使用了七个模型 GeResult_ensemble_TTA.py，使用TTA集成方法对模型进行集成

dataloader文件夹中是数据集的制作

Name		Name	Last commit message	Last commit date
Latest commit History 6 Commits
Dataloader		Dataloader
basenet		basenet
1_train.txt		1_train.txt
1_val.txt		1_val.txt
GeMergeResult.py		GeMergeResult.py
GeResult.py		GeResult.py
GeResult_ensemble.py		GeResult_ensemble.py
GeResult_ensemble_TTA.py		GeResult_ensemble_TTA.py
GeResult_ensemble_selecttotrain.py		GeResult_ensemble_selecttotrain.py
MM_val.py		MM_val.py
MM_val1.py		MM_val1.py
MM_val2.py		MM_val2.py
MM_val_ensemble.py		MM_val_ensemble.py
README.md		README.md
all.txt		all.txt
multimodal_train_val.py		multimodal_train_val.py
multimodal_train_val1.py		multimodal_train_val1.py
multimodal_train_val2.py		multimodal_train_val2.py
multimodal_train_val3.py		multimodal_train_val3.py
val.py		val.py
val_TTA.py		val_TTA.py
val_TTA_MM.py		val_TTA_MM.py
visit2array_adapt.py		visit2array_adapt.py

Provide feedback