From 69c160404a8c32638e9bb8eb65702c659c9f51a5 Mon Sep 17 00:00:00 2001 From: Bobholamovic Date: Mon, 12 Sep 2022 14:21:02 +0800 Subject: [PATCH] Add seg models --- docs/intro/model_zoo.md | 23 +++-- paddlers/tasks/segmenter.py | 11 ++- test_tipc/configs/cd/bit/bit.yaml | 8 -- .../configs/cd/changeformer/changeformer.yaml | 8 -- test_tipc/configs/cd/fccdn/fccdn.yaml | 13 --- .../seg/bisenetv2/bisenetv2_rsseg.yaml | 11 +++ .../seg/bisenetv2/train_infer_python.txt | 53 +++++++++++ .../seg/fast_scnn/fast_scnn_rsseg.yaml | 11 +++ .../seg/fast_scnn/train_infer_python.txt | 53 +++++++++++ test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml | 11 +++ .../configs/seg/hrnet/train_infer_python.txt | 53 +++++++++++ test_tipc/docs/test_train_inference_python.md | 8 ++ tutorials/train/README.md | 4 + .../train/semantic_segmentation/bisenetv2.py | 93 +++++++++++++++++++ .../train/semantic_segmentation/fast_scnn.py | 93 +++++++++++++++++++ .../train/semantic_segmentation/hrnet.py | 93 +++++++++++++++++++ 16 files changed, 504 insertions(+), 42 deletions(-) delete mode 100644 test_tipc/configs/cd/bit/bit.yaml delete mode 100644 test_tipc/configs/cd/changeformer/changeformer.yaml delete mode 100644 test_tipc/configs/cd/fccdn/fccdn.yaml create mode 100644 test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml create mode 100644 test_tipc/configs/seg/bisenetv2/train_infer_python.txt create mode 100644 test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml create mode 100644 test_tipc/configs/seg/fast_scnn/train_infer_python.txt create mode 100644 test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml create mode 100644 test_tipc/configs/seg/hrnet/train_infer_python.txt create mode 100644 tutorials/train/semantic_segmentation/bisenetv2.py create mode 100644 tutorials/train/semantic_segmentation/fast_scnn.py create mode 100644 tutorials/train/semantic_segmentation/hrnet.py diff --git a/docs/intro/model_zoo.md b/docs/intro/model_zoo.md index 9281df8..f8b520d 100644 --- a/docs/intro/model_zoo.md +++ b/docs/intro/model_zoo.md @@ -20,18 +20,21 @@ PaddleRS目前已支持的全部模型如下（标注\*的为遥感专用模型 | 变化检测 | \*FCCDN | 是 | | 变化检测 | \*SNUNet | 是 | | 变化检测 | \*STANet | 是 | -| 场景分类 | CondenseNetV2 | 是 | -| 场景分类 | HRNet | 是 | -| 场景分类 | MobileNetV3 | 是 | -| 场景分类 | ResNet50-vd | 是 | +| 场景分类 | CondenseNet V2 | 是 | +| 场景分类 | HRNet | 否 | +| 场景分类 | MobileNetV3 | 否 | +| 场景分类 | ResNet50-vd | 否 | | 图像复原 | DRN | 否 | | 图像复原 | ESRGAN | 否 | | 图像复原 | LESRCNN | 否 | -| 目标检测 | Faster R-CNN | 是 | -| 目标检测 | PP-YOLO | 是 | -| 目标检测 | PP-YOLO Tiny | 是 | -| 目标检测 | PP-YOLOv2 | 是 | -| 目标检测 | YOLOv3 | 是 | +| 目标检测 | Faster R-CNN | 否 | +| 目标检测 | PP-YOLO | 否 | +| 目标检测 | PP-YOLO Tiny | 否 | +| 目标检测 | PP-YOLOv2 | 否 | +| 目标检测 | YOLOv3 | 否 | +| 图像分割 | BiSeNet V2 | 是 | | 图像分割 | DeepLab V3+ | 是 | -| 图像分割 | \*FarSeg | 否 | +| 图像分割 | \*FarSeg | 是 | +| 图像分割 | Fast-SCNN | 是 | +| 图像分割 | HRNet | 是 | | 图像分割 | UNet | 是 | diff --git a/paddlers/tasks/segmenter.py b/paddlers/tasks/segmenter.py index 83cbffa..9e6b584 100644 --- a/paddlers/tasks/segmenter.py +++ b/paddlers/tasks/segmenter.py @@ -806,7 +806,7 @@ class UNet(BaseSegmenter): }) super(UNet, self).__init__( model_name='UNet', - input_channel=in_channels, + in_channels=in_channels, num_classes=num_classes, use_mixed_loss=use_mixed_loss, losses=losses, @@ -834,7 +834,7 @@ class DeepLabV3P(BaseSegmenter): if params.get('with_net', True): with DisablePrint(): backbone = getattr(ppseg.models, backbone)( - input_channel=in_channels, output_stride=output_stride) + in_channels=in_channels, output_stride=output_stride) else: backbone = None params.update({ @@ -854,6 +854,7 @@ class DeepLabV3P(BaseSegmenter): class FastSCNN(BaseSegmenter): def __init__(self, + in_channels=3, num_classes=2, use_mixed_loss=False, losses=None, @@ -862,6 +863,7 @@ class FastSCNN(BaseSegmenter): params.update({'align_corners': align_corners}) super(FastSCNN, self).__init__( model_name='FastSCNN', + in_channels=in_channels, num_classes=num_classes, use_mixed_loss=use_mixed_loss, losses=losses, @@ -870,6 +872,7 @@ class FastSCNN(BaseSegmenter): class HRNet(BaseSegmenter): def __init__(self, + in_channels=3, num_classes=2, width=48, use_mixed_loss=False, @@ -884,7 +887,7 @@ class HRNet(BaseSegmenter): if params.get('with_net', True): with DisablePrint(): backbone = getattr(ppseg.models, self.backbone_name)( - align_corners=align_corners) + in_channels=in_channels, align_corners=align_corners) else: backbone = None @@ -900,6 +903,7 @@ class HRNet(BaseSegmenter): class BiSeNetV2(BaseSegmenter): def __init__(self, + in_channels=3, num_classes=2, use_mixed_loss=False, losses=None, @@ -908,6 +912,7 @@ class BiSeNetV2(BaseSegmenter): params.update({'align_corners': align_corners}) super(BiSeNetV2, self).__init__( model_name='BiSeNetV2', + in_channels=in_channels, num_classes=num_classes, use_mixed_loss=use_mixed_loss, losses=losses, diff --git a/test_tipc/configs/cd/bit/bit.yaml b/test_tipc/configs/cd/bit/bit.yaml deleted file mode 100644 index 3d3c62b..0000000 --- a/test_tipc/configs/cd/bit/bit.yaml +++ /dev/null @@ -1,8 +0,0 @@ -# Basic configurations of BIT - -_base_: ../_base_/airchange.yaml - -save_dir: ./test_tipc/output/cd/bit/ - -model: !Node - type: BIT \ No newline at end of file diff --git a/test_tipc/configs/cd/changeformer/changeformer.yaml b/test_tipc/configs/cd/changeformer/changeformer.yaml deleted file mode 100644 index 785749d..0000000 --- a/test_tipc/configs/cd/changeformer/changeformer.yaml +++ /dev/null @@ -1,8 +0,0 @@ -# Basic configurations of ChangeFormer - -_base_: ../_base_/airchange.yaml - -save_dir: ./test_tipc/output/cd/changeformer/ - -model: !Node - type: ChangeFormer \ No newline at end of file diff --git a/test_tipc/configs/cd/fccdn/fccdn.yaml b/test_tipc/configs/cd/fccdn/fccdn.yaml deleted file mode 100644 index 8b93717..0000000 --- a/test_tipc/configs/cd/fccdn/fccdn.yaml +++ /dev/null @@ -1,13 +0,0 @@ -# Basic configurations of FCCDN - -_base_: ../_base_/airchange.yaml - -save_dir: ./test_tipc/output/cd/fccdn/ - -model: !Node - type: FCCDN - -learning_rate: 0.07 -lr_decay_power: 0.6 -log_interval_steps: 100 -save_interval_epochs: 3 diff --git a/test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml b/test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml new file mode 100644 index 0000000..81c29b6 --- /dev/null +++ b/test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml @@ -0,0 +1,11 @@ +# Configurations of BiSeNet V2 with RSSeg dataset + +_base_: ../_base_/rsseg.yaml + +save_dir: ./test_tipc/output/seg/bisenetv2/ + +model: !Node + type: BiSeNet V2 + args: + in_channels: 10 + num_classes: 5 \ No newline at end of file diff --git a/test_tipc/configs/seg/bisenetv2/train_infer_python.txt b/test_tipc/configs/seg/bisenetv2/train_infer_python.txt new file mode 100644 index 0000000..2fcad98 --- /dev/null +++ b/test_tipc/configs/seg/bisenetv2/train_infer_python.txt @@ -0,0 +1,53 @@ +===========================train_params=========================== +model_name:seg:bisenetv2 +python:python +gpu_list:0|0,1 +use_gpu:null|null +--precision:null +--num_epochs:lite_train_lite_infer=3|lite_train_whole_infer=3|whole_train_whole_infer=20 +--save_dir:adaptive +--train_batch_size:lite_train_lite_infer=4|lite_train_whole_infer=4|whole_train_whole_infer=4 +--model_path:null +--config:lite_train_lite_infer=./test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml|lite_train_whole_infer=./test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml|whole_train_whole_infer=./test_tipc/configs/seg/bisenetv2/bisenetv2_rsseg.yaml +train_model_name:best_model +null:null +## +trainer:norm +norm_train:test_tipc/run_task.py train seg +pact_train:null +fpgm_train:null +distill_train:null +null:null +null:null +## +===========================eval_params=========================== +eval:null +null:null +## +===========================export_params=========================== +--save_dir:adaptive +--model_dir:adaptive +--fixed_input_shape:[-1,10,512,512] +norm_export:deploy/export/export_model.py +quant_export:null +fpgm_export:null +distill_export:null +export1:null +export2:null +===========================infer_params=========================== +infer_model:null +infer_export:null +infer_quant:False +inference:test_tipc/infer.py +--device:cpu|gpu +--enable_mkldnn:True +--cpu_threads:6 +--batch_size:1 +--use_trt:False +--precision:fp32 +--model_dir:null +--config:null +--save_log_path:null +--benchmark:True +--model_name:bisenetv2 +null:null \ No newline at end of file diff --git a/test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml b/test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml new file mode 100644 index 0000000..145ce47 --- /dev/null +++ b/test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml @@ -0,0 +1,11 @@ +# Configurations of Fast-SCNN with RSSeg dataset + +_base_: ../_base_/rsseg.yaml + +save_dir: ./test_tipc/output/seg/fast_scnn/ + +model: !Node + type: Fast-SCNN + args: + in_channels: 10 + num_classes: 5 \ No newline at end of file diff --git a/test_tipc/configs/seg/fast_scnn/train_infer_python.txt b/test_tipc/configs/seg/fast_scnn/train_infer_python.txt new file mode 100644 index 0000000..f5bdcd5 --- /dev/null +++ b/test_tipc/configs/seg/fast_scnn/train_infer_python.txt @@ -0,0 +1,53 @@ +===========================train_params=========================== +model_name:seg:fast_scnn +python:python +gpu_list:0|0,1 +use_gpu:null|null +--precision:null +--num_epochs:lite_train_lite_infer=3|lite_train_whole_infer=3|whole_train_whole_infer=20 +--save_dir:adaptive +--train_batch_size:lite_train_lite_infer=4|lite_train_whole_infer=4|whole_train_whole_infer=4 +--model_path:null +--config:lite_train_lite_infer=./test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml|lite_train_whole_infer=./test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml|whole_train_whole_infer=./test_tipc/configs/seg/fast_scnn/fast_scnn_rsseg.yaml +train_model_name:best_model +null:null +## +trainer:norm +norm_train:test_tipc/run_task.py train seg +pact_train:null +fpgm_train:null +distill_train:null +null:null +null:null +## +===========================eval_params=========================== +eval:null +null:null +## +===========================export_params=========================== +--save_dir:adaptive +--model_dir:adaptive +--fixed_input_shape:[-1,10,512,512] +norm_export:deploy/export/export_model.py +quant_export:null +fpgm_export:null +distill_export:null +export1:null +export2:null +===========================infer_params=========================== +infer_model:null +infer_export:null +infer_quant:False +inference:test_tipc/infer.py +--device:cpu|gpu +--enable_mkldnn:True +--cpu_threads:6 +--batch_size:1 +--use_trt:False +--precision:fp32 +--model_dir:null +--config:null +--save_log_path:null +--benchmark:True +--model_name:fast_scnn +null:null \ No newline at end of file diff --git a/test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml b/test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml new file mode 100644 index 0000000..a7eff81 --- /dev/null +++ b/test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml @@ -0,0 +1,11 @@ +# Configurations of HRNet with RSSeg dataset + +_base_: ../_base_/rsseg.yaml + +save_dir: ./test_tipc/output/seg/hrnet/ + +model: !Node + type: HRNet + args: + in_channels: 10 + num_classes: 5 \ No newline at end of file diff --git a/test_tipc/configs/seg/hrnet/train_infer_python.txt b/test_tipc/configs/seg/hrnet/train_infer_python.txt new file mode 100644 index 0000000..d1cafdc --- /dev/null +++ b/test_tipc/configs/seg/hrnet/train_infer_python.txt @@ -0,0 +1,53 @@ +===========================train_params=========================== +model_name:seg:hrnet +python:python +gpu_list:0|0,1 +use_gpu:null|null +--precision:null +--num_epochs:lite_train_lite_infer=3|lite_train_whole_infer=3|whole_train_whole_infer=20 +--save_dir:adaptive +--train_batch_size:lite_train_lite_infer=4|lite_train_whole_infer=4|whole_train_whole_infer=4 +--model_path:null +--config:lite_train_lite_infer=./test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml|lite_train_whole_infer=./test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml|whole_train_whole_infer=./test_tipc/configs/seg/hrnet/hrnet_rsseg.yaml +train_model_name:best_model +null:null +## +trainer:norm +norm_train:test_tipc/run_task.py train seg +pact_train:null +fpgm_train:null +distill_train:null +null:null +null:null +## +===========================eval_params=========================== +eval:null +null:null +## +===========================export_params=========================== +--save_dir:adaptive +--model_dir:adaptive +--fixed_input_shape:[-1,10,512,512] +norm_export:deploy/export/export_model.py +quant_export:null +fpgm_export:null +distill_export:null +export1:null +export2:null +===========================infer_params=========================== +infer_model:null +infer_export:null +infer_quant:False +inference:test_tipc/infer.py +--device:cpu|gpu +--enable_mkldnn:True +--cpu_threads:6 +--batch_size:1 +--use_trt:False +--precision:fp32 +--model_dir:null +--config:null +--save_log_path:null +--benchmark:True +--model_name:hrnet +null:null \ No newline at end of file diff --git a/test_tipc/docs/test_train_inference_python.md b/test_tipc/docs/test_train_inference_python.md index b117cce..2d01ca3 100644 --- a/test_tipc/docs/test_train_inference_python.md +++ b/test_tipc/docs/test_train_inference_python.md @@ -19,6 +19,7 @@ Linux GPU/CPU 基础训练推理测试的主程序为`test_train_inference_pytho | 变化检测 | FC-Siam-conc | 正常训练 | 正常训练 | IoU=65.79% | | 变化检测 | FC-Siam-diff | 正常训练 | 正常训练 | IoU=61.23% | | 变化检测 | FCCDN | 正常训练 | 正常训练 | IoU=24.42% | +| 场景分类 | CondenseNet V2 | 正常训练 | 正常训练 | Acc(top1)= | | 场景分类 | HRNet | 正常训练 | 正常训练 | Acc(top1)=99.37% | | 场景分类 | MobileNetV3 | 正常训练 | 正常训练 | Acc(top1)=99.58% | | 场景分类 | ResNet50-vd | 正常训练 | 正常训练 | Acc(top1)=99.26% | @@ -30,8 +31,11 @@ Linux GPU/CPU 基础训练推理测试的主程序为`test_train_inference_pytho | 目标检测 | PP-YOLO Tiny | 正常训练 | 正常训练 | mAP=44.27% | | 目标检测 | PP-YOLOv2 | 正常训练 | 正常训练 | mAP=59.37% | | 目标检测 | YOLOv3 | 正常训练 | 正常训练 | mAP=47.33% | +| 图像分割 | BiSeNet V2 | 正常训练 | 正常训练 | mIoU= | | 图像分割 | DeepLab V3+ | 正常训练 | 正常训练 | mIoU=56.05% | | 图像分割 | FarSeg | 正常训练 | 正常训练 | mIoU=49.58% | +| 图像分割 | Fast-SCNN | 正常训练 | 正常训练 | mIoU= | +| 图像分割 | HRNet | 正常训练 | 正常训练 | mIoU= | | 图像分割 | UNet | 正常训练 | 正常训练 | mIoU=55.50% | *注：参考预测精度为whole_train_whole_infer模式下单卡训练汇报的精度数据。* @@ -50,6 +54,7 @@ Linux GPU/CPU 基础训练推理测试的主程序为`test_train_inference_pytho | 变化检测 | FC-EF | 支持 | 支持 | 1 | | 变化检测 | FC-Siam-conc | 支持 | 支持 | 1 | | 变化检测 | FC-Siam-diff | 支持 | 支持 | 1 | +| 场景分类 | CondenseNet V2 | 支持 | 支持 | 1 | | 场景分类 | HRNet | 支持 | 支持 | 1 | | 场景分类 | MobileNetV3 | 支持 | 支持 | 1 | | 场景分类 | ResNet50-vd | 支持 | 支持 | 1 | @@ -61,8 +66,11 @@ Linux GPU/CPU 基础训练推理测试的主程序为`test_train_inference_pytho | 目标检测 | PP-YOLO Tiny | 支持 | 支持 | 1 | | 目标检测 | PP-YOLOv2 | 支持 | 支持 | 1 | | 目标检测 | YOLOv3 | 支持 | 支持 | 1 | +| 图像分割 | BiSeNet V2 | 支持 | 支持 | 1 | | 图像分割 | DeepLab V3+ | 支持 | 支持 | 1 | | 图像分割 | FarSeg | 支持 | 支持 | 1 | +| 图像分割 | Fast-SCNN | 支持 | 支持 | 1 | +| 图像分割 | HRNet | 支持 | 支持 | 1 | | 图像分割 | UNet | 支持 | 支持 | 1 | ## 2 测试流程 diff --git a/tutorials/train/README.md b/tutorials/train/README.md index 44c2491..105f98a 100644 --- a/tutorials/train/README.md +++ b/tutorials/train/README.md @@ -15,6 +15,7 @@ |change_detection/fccdn.py | 变化检测 | FCCDN | |change_detection/snunet.py | 变化检测 | SNUNet | |change_detection/stanet.py | 变化检测 | STANet | +|classification/condensenetv2.py | 场景分类 | CondenseNet V2 | |classification/hrnet.py | 场景分类 | HRNet | |classification/mobilenetv3.py | 场景分类 | MobileNetV3 | |classification/resnet50_vd.py | 场景分类 | ResNet50-vd | @@ -26,8 +27,11 @@ |object_detection/ppyolo_tiny.py | 目标检测 | PP-YOLO Tiny | |object_detection/ppyolov2.py | 目标检测 | PP-YOLOv2 | |object_detection/yolov3.py | 目标检测 | YOLOv3 | +|semantic_segmentation/bisenetv2.py | 图像分割 | BiSeNet V2 | |semantic_segmentation/deeplabv3p.py | 图像分割 | DeepLab V3+ | |semantic_segmentation/farseg.py | 图像分割 | FarSeg | +|semantic_segmentation/fast_scnn.py | 图像分割 | Fast-SCNN | +|semantic_segmentation/hrnet.py | 图像分割 | HRNet | |semantic_segmentation/unet.py | 图像分割 | UNet | ## 环境准备 diff --git a/tutorials/train/semantic_segmentation/bisenetv2.py b/tutorials/train/semantic_segmentation/bisenetv2.py new file mode 100644 index 0000000..6d2754d --- /dev/null +++ b/tutorials/train/semantic_segmentation/bisenetv2.py @@ -0,0 +1,93 @@ +#!/usr/bin/env python + +# 图像分割模型BiSeNet V2训练示例脚本 +# 执行此脚本前，请确认已正确安装PaddleRS库 + +import paddlers as pdrs +from paddlers import transforms as T + +# 数据集存放目录 +DATA_DIR = './data/rsseg/' +# 训练集`file_list`文件路径 +TRAIN_FILE_LIST_PATH = './data/rsseg/train.txt' +# 验证集`file_list`文件路径 +EVAL_FILE_LIST_PATH = './data/rsseg/val.txt' +# 数据集类别信息文件路径 +LABEL_LIST_PATH = './data/rsseg/labels.txt' +# 实验目录，保存输出的模型权重和结果 +EXP_DIR = './output/unet/' + +# 影像波段数量 +NUM_BANDS = 10 + +# 下载和解压多光谱地块分类数据集 +pdrs.utils.download_and_decompress( + 'https://paddlers.bj.bcebos.com/datasets/rsseg.zip', path='./data/') + +# 定义训练和验证时使用的数据变换（数据增强、预处理等） +# 使用Compose组合多种变换方式。Compose中包含的变换将按顺序串行执行 +# API说明：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/apis/data.md +train_transforms = T.Compose([ + # 读取影像 + T.DecodeImg(), + # 将影像缩放到512x512大小 + T.Resize(target_size=512), + # 以50%的概率实施随机水平翻转 + T.RandomHorizontalFlip(prob=0.5), + # 将数据归一化到[-1,1] + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ArrangeSegmenter('train') +]) + +eval_transforms = T.Compose([ + T.DecodeImg(), + T.Resize(target_size=512), + # 验证阶段与训练阶段的数据归一化方式必须相同 + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ReloadMask(), + T.ArrangeSegmenter('eval') +]) + +# 分别构建训练和验证所用的数据集 +train_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=TRAIN_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=train_transforms, + num_workers=0, + shuffle=True) + +eval_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=EVAL_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=eval_transforms, + num_workers=0, + shuffle=False) + +# 构建BiSeNet V2模型 +# 目前已支持的模型请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/intro/model_zoo.md +# 模型输入参数请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/paddlers/tasks/segmenter.py +model = pdrs.tasks.seg.BiSeNetV2( + in_channels=NUM_BANDS, num_classes=len(train_dataset.labels)) + +# 执行模型训练 +model.train( + num_epochs=10, + train_dataset=train_dataset, + train_batch_size=4, + eval_dataset=eval_dataset, + save_interval_epochs=5, + # 每多少次迭代记录一次日志 + log_interval_steps=4, + save_dir=EXP_DIR, + # 初始学习率大小 + learning_rate=0.001, + # 是否使用early stopping策略，当精度不再改善时提前终止训练 + early_stop=False, + # 是否启用VisualDL日志功能 + use_vdl=True, + # 指定从某个检查点继续训练 + resume_checkpoint=None) diff --git a/tutorials/train/semantic_segmentation/fast_scnn.py b/tutorials/train/semantic_segmentation/fast_scnn.py new file mode 100644 index 0000000..df15ece --- /dev/null +++ b/tutorials/train/semantic_segmentation/fast_scnn.py @@ -0,0 +1,93 @@ +#!/usr/bin/env python + +# 图像分割模型Fast-SCNN训练示例脚本 +# 执行此脚本前，请确认已正确安装PaddleRS库 + +import paddlers as pdrs +from paddlers import transforms as T + +# 数据集存放目录 +DATA_DIR = './data/rsseg/' +# 训练集`file_list`文件路径 +TRAIN_FILE_LIST_PATH = './data/rsseg/train.txt' +# 验证集`file_list`文件路径 +EVAL_FILE_LIST_PATH = './data/rsseg/val.txt' +# 数据集类别信息文件路径 +LABEL_LIST_PATH = './data/rsseg/labels.txt' +# 实验目录，保存输出的模型权重和结果 +EXP_DIR = './output/unet/' + +# 影像波段数量 +NUM_BANDS = 10 + +# 下载和解压多光谱地块分类数据集 +pdrs.utils.download_and_decompress( + 'https://paddlers.bj.bcebos.com/datasets/rsseg.zip', path='./data/') + +# 定义训练和验证时使用的数据变换（数据增强、预处理等） +# 使用Compose组合多种变换方式。Compose中包含的变换将按顺序串行执行 +# API说明：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/apis/data.md +train_transforms = T.Compose([ + # 读取影像 + T.DecodeImg(), + # 将影像缩放到512x512大小 + T.Resize(target_size=512), + # 以50%的概率实施随机水平翻转 + T.RandomHorizontalFlip(prob=0.5), + # 将数据归一化到[-1,1] + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ArrangeSegmenter('train') +]) + +eval_transforms = T.Compose([ + T.DecodeImg(), + T.Resize(target_size=512), + # 验证阶段与训练阶段的数据归一化方式必须相同 + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ReloadMask(), + T.ArrangeSegmenter('eval') +]) + +# 分别构建训练和验证所用的数据集 +train_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=TRAIN_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=train_transforms, + num_workers=0, + shuffle=True) + +eval_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=EVAL_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=eval_transforms, + num_workers=0, + shuffle=False) + +# 构建Fast-SCNN模型 +# 目前已支持的模型请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/intro/model_zoo.md +# 模型输入参数请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/paddlers/tasks/segmenter.py +model = pdrs.tasks.seg.FastSCNN( + in_channels=NUM_BANDS, num_classes=len(train_dataset.labels)) + +# 执行模型训练 +model.train( + num_epochs=10, + train_dataset=train_dataset, + train_batch_size=4, + eval_dataset=eval_dataset, + save_interval_epochs=5, + # 每多少次迭代记录一次日志 + log_interval_steps=4, + save_dir=EXP_DIR, + # 初始学习率大小 + learning_rate=0.001, + # 是否使用early stopping策略，当精度不再改善时提前终止训练 + early_stop=False, + # 是否启用VisualDL日志功能 + use_vdl=True, + # 指定从某个检查点继续训练 + resume_checkpoint=None) diff --git a/tutorials/train/semantic_segmentation/hrnet.py b/tutorials/train/semantic_segmentation/hrnet.py new file mode 100644 index 0000000..e8a9cdb --- /dev/null +++ b/tutorials/train/semantic_segmentation/hrnet.py @@ -0,0 +1,93 @@ +#!/usr/bin/env python + +# 图像分割模型HRNet训练示例脚本 +# 执行此脚本前，请确认已正确安装PaddleRS库 + +import paddlers as pdrs +from paddlers import transforms as T + +# 数据集存放目录 +DATA_DIR = './data/rsseg/' +# 训练集`file_list`文件路径 +TRAIN_FILE_LIST_PATH = './data/rsseg/train.txt' +# 验证集`file_list`文件路径 +EVAL_FILE_LIST_PATH = './data/rsseg/val.txt' +# 数据集类别信息文件路径 +LABEL_LIST_PATH = './data/rsseg/labels.txt' +# 实验目录，保存输出的模型权重和结果 +EXP_DIR = './output/unet/' + +# 影像波段数量 +NUM_BANDS = 10 + +# 下载和解压多光谱地块分类数据集 +pdrs.utils.download_and_decompress( + 'https://paddlers.bj.bcebos.com/datasets/rsseg.zip', path='./data/') + +# 定义训练和验证时使用的数据变换（数据增强、预处理等） +# 使用Compose组合多种变换方式。Compose中包含的变换将按顺序串行执行 +# API说明：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/apis/data.md +train_transforms = T.Compose([ + # 读取影像 + T.DecodeImg(), + # 将影像缩放到512x512大小 + T.Resize(target_size=512), + # 以50%的概率实施随机水平翻转 + T.RandomHorizontalFlip(prob=0.5), + # 将数据归一化到[-1,1] + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ArrangeSegmenter('train') +]) + +eval_transforms = T.Compose([ + T.DecodeImg(), + T.Resize(target_size=512), + # 验证阶段与训练阶段的数据归一化方式必须相同 + T.Normalize( + mean=[0.5] * NUM_BANDS, std=[0.5] * NUM_BANDS), + T.ReloadMask(), + T.ArrangeSegmenter('eval') +]) + +# 分别构建训练和验证所用的数据集 +train_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=TRAIN_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=train_transforms, + num_workers=0, + shuffle=True) + +eval_dataset = pdrs.datasets.SegDataset( + data_dir=DATA_DIR, + file_list=EVAL_FILE_LIST_PATH, + label_list=LABEL_LIST_PATH, + transforms=eval_transforms, + num_workers=0, + shuffle=False) + +# 构建HRNet模型 +# 目前已支持的模型请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/docs/intro/model_zoo.md +# 模型输入参数请参考：https://github.com/PaddlePaddle/PaddleRS/blob/develop/paddlers/tasks/segmenter.py +model = pdrs.tasks.seg.HRNet( + in_channels=NUM_BANDS, num_classes=len(train_dataset.labels)) + +# 执行模型训练 +model.train( + num_epochs=10, + train_dataset=train_dataset, + train_batch_size=4, + eval_dataset=eval_dataset, + save_interval_epochs=5, + # 每多少次迭代记录一次日志 + log_interval_steps=4, + save_dir=EXP_DIR, + # 初始学习率大小 + learning_rate=0.001, + # 是否使用early stopping策略，当精度不再改善时提前终止训练 + early_stop=False, + # 是否启用VisualDL日志功能 + use_vdl=True, + # 指定从某个检查点继续训练 + resume_checkpoint=None)