MASK_RCNN_Resnet50(Pytorch版本)

配置环境

1. OS 名称及版本: Microsoft Windows 10 专业版 10.0.17763 暂缺 Build 17763

2. CUDA版本¹： cuda_11.0.1_451.22_win10.exe

3. Python版本： Anaconda3-2020.02-Windows-x86_64.exe

4. Pytorch版本： torch-1.5.0-cp37-cp37m-win_amd64.whl

5. Torchvision版本： torchvision-0.6.0-cp37-cp37m-win_amd64.whl

6. IDE名称及版本： pycharm-community-2020.1.1.exe

7. 文档制作软件： typora-setup-x64.exe

8. 浏览器： QQBrowser_Setup_Qqpcmgr_10.5.4043.400.exe

9. 代码管理工具： Git-2.27.0-64-bit.exe

10. 图片标注工具: labelme Version: 4.4.0 (pip方式安装)

包国内源

（1）阿里云 http://mirrors.aliyun.com/pypi/simple/
（2）豆瓣http://pypi.douban.com/simple/
（3）清华大学 https://pypi.tuna.tsinghua.edu.cn/simple/
（4）中国科学技术大学 http://pypi.mirrors.ustc.edu.cn/simple/
（5）华中科技大学http://pypi.hustunique.com/

安装文件

1. 安装coco的api [cocoapi]：主要用到其中的IOU计算的库来评价模型的性能。

git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
python setup.py build_ext install

2. 数据集 [PennFudanPed]：本教程使用Penn-Fudan的行人检测和分割数据集来训练Mask R-CNN实例分割模型。Penn-Fudan数据集中有170张图像，包含345个行人的实例。图像中场景主要是校园和城市街景，每张图中至少有一个行人，具体的介绍和下载地址如下:

# 下载Penn-Fudan dataset
wget https://www.cis.upenn.edu/~jshi/ped_html/PennFudanPed.zip
# 解压到当前目录
unzip PennFudanPed.zip

3. mask_rcnn文件

# 网络模型
maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth #预训练的maskrcnn
resnet50-19c8e357.pth                   #没训练的maskrcnn

cd C:\Users\Administrator\.cache\torch\checkpoints
# 将网络模型放入此文件夹
copy {网络模型} 

4. 工具文件[vision-master]：在PyTorch官方的references/detection/中，有一些封装好的用于模型训练和测试的函数，其中references/detection/engine.py、references/detection/utils.py、references/detection/transforms.py是我们需要用到的。首先，将这些文件拷贝过来。

# Download TorchVision repo to use some files from references/detection
git clone https://github.com/pytorch/vision.git
cd visiongit checkout v0.4.0 
cp references/detection/utils.py ../
cp references/detection/transforms.py ../
cp references/detection/coco_eval.py ../
cp references/detection/engine.py ../
cp references/detection/coco_utils.py ../

​

数据批处理

# coding:utf-8
__author__ = "Xuyc"
__date__ = "2020/6/19 11:11"
__filename__ = "jsontool.py"

import os

os.chdir(r'..\json')
print(os.getcwd())
for i in os.listdir(r'..\json'):
    print(i)
    os.system(r'python ..\Anaconda3\Lib\site-packages\labelme\cli\json_to_dataset.py "{}"'.format(i))
if not os.path.exists('img'):
    os.mkdir('img')
if not os.path.exists('mask'):
    os.mkdir('mask')
dirs = os.listdir(r'..\json')
jsons = []
for i in os.listdir(r'..\json'):
    if '_json' in i:
        jsons.append(i)
print(jsons)
for i in jsons:
    os.system(r'copy "{}\img.png" "img\{}.png"'.format(i, i))
    os.system(r'copy "{}\label.png" "mask\{}_mask.png"'.format(i, i))

采坑记录

1. windows安装pycocotools时报错，如下：

cl: 命令行 error D8021 :无效的数值参数“/Wno-cpp”

解决方案：

1. 打开coco\PythonAPI目录下的 setup.py文件，修改ext_modules extra_compile_args=[’-Wno-cpp’, ‘-Wno-unused-function’, ‘-std=c99’]，如下：

from setuptools import setup, Extension
import numpy as np

# To compile and install locally run "python setup.py build_ext --inplace"
# To install library to Python site-packages run "python setup.py build_ext install"

ext_modules = [
　　Extension(
　　　　'pycocotools._mask',
　　　　sources=['../common/maskApi.c', 'pycocotools/_mask.pyx'],
　　　　include_dirs = [np.get_include(), '../common'],

　　　　#修改位置，去掉
　　　　#extra_compile_args=['-Wno-cpp', '-Wno-unused-function', '-std=c99'],
　　　　extra_compile_args=['', '', ''],
　　)
]

setup(
　　name='pycocotools',
　　packages=['pycocotools'],
　　package_dir = {'pycocotools': 'pycocotools'},
　　install_requires=[
　　　　'setuptools>=18.0',
　　　　'cython>=0.27.3',
　　　　'matplotlib>=2.1.0'
　　],
　　version='2.0',
　　ext_modules= ext_modules
)

​ 然后保存setup.py文件，执行。

2. 在测试模型性能的时候，如果出现ValueError: Does not understand character buffer dtype format string (’?’)：

File "build/bdist.linux-x86_64/egg/pycocotools/mask.py", line 82, in encode  File "pycocotools/_mask.pyx", line 137, in pycocotools._mask.encodeValueError: Does not understand character buffer dtype format string ('?)

​ 通过修改coco_eval.py中mask_util.encode一行，添加dtype=np.uint8，即可搞定.

In coco_eval.py:

rles = [
mask_util.encode(np.array(mask[0, :, :, np.newaxis], dtype=np.uint8, order="F"))[0]
for mask in masks
]

3 缺少C++ 14.0
官网下载，或是评论区留言

程序源码

# Sample code from the TorchVision 0.3 Object Detection Finetuning Tutorial
# http://pytorch.org/tutorials/intermediate/torchvision_tutorial.html

__author__ = "torch"
__filename__ = "tv-training-code.py"

import os
import numpy as np
import torch
import torch.utils.data
from PIL import Image

import torchvision
from torchvision.models.detection.faster_rcnn import FastRCNNPredictor
from torchvision.models.detection.mask_rcnn import MaskRCNNPredictor

from engine import train_one_epoch, evaluate
import utils
import transforms as T


class PennFudanDataset(object):
    def __init__(self, root, transforms):
        self.root = root
        self.transforms = transforms
        # load all image files, sorting them to
        # ensure that they are aligned
        self.imgs = list(sorted(os.listdir(os.path.join(root, "PNGImages"))))
        self.masks = list(sorted(os.listdir(os.path.join(root, "PedMasks"))))

    def __getitem__(self, idx):
        # load images ad masks
        img_path = os.path.join(self.root, "PNGImages", self.imgs[idx])
        mask_path = os.path.join(self.root, "PedMasks", self.masks[idx])
        img = Image.open(img_path).convert("RGB")
        # note that we haven't converted the mask to RGB,
        # because each color corresponds to a different instance
        # with 0 being background
        mask = Image.open(mask_path)

        mask = np.array(mask)
        # instances are encoded as different colors
        obj_ids = np.unique(mask)
        # first id is the background, so remove it
        obj_ids = obj_ids[1:]

        # split the color-encoded mask into a set
        # of binary masks
        masks = mask == obj_ids[:, None, None]

        # get bounding box coordinates for each mask
        num_objs = len(obj_ids)
        boxes = []
        for i in range(num_objs):
            pos = np.where(masks[i])
            xmin = np.min(pos[1])
            xmax = np.max(pos[1])
            ymin = np.min(pos[0])
            ymax = np.max(pos[0])
            boxes.append([xmin, ymin, xmax, ymax])

        boxes = torch.as_tensor(boxes, dtype=torch.float32)
        # there is only one class
        labels = torch.ones((num_objs,), dtype=torch.int64)
        masks = torch.as_tensor(masks, dtype=torch.uint8)

        image_id = torch.tensor([idx])
        area = (boxes[:, 3] - boxes[:, 1]) * (boxes[:, 2] - boxes[:, 0])
        # suppose all instances are not crowd
        iscrowd = torch.zeros((num_objs,), dtype=torch.int64)

        target = {}
        target["boxes"] = boxes
        target["labels"] = labels
        target["masks"] = masks
        target["image_id"] = image_id
        target["area"] = area
        target["iscrowd"] = iscrowd

        if self.transforms is not None:
            img, target = self.transforms(img, target)

        return img, target

    def __len__(self):
        return len(self.imgs)


def get_model_instance_segmentation(num_classes):
    # load an instance segmentation model pre-trained pre-trained on COCO
    model = torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True)

    # get number of input features for the classifier
    in_features = model.roi_heads.box_predictor.cls_score.in_features
    # replace the pre-trained head with a new one
    model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)

    # now get the number of input features for the mask classifier
    in_features_mask = model.roi_heads.mask_predictor.conv5_mask.in_channels
    hidden_layer = 256
    # and replace the mask predictor with a new one
    model.roi_heads.mask_predictor = MaskRCNNPredictor(in_features_mask,
                                                       hidden_layer,
                                                       num_classes)

    return model


def get_transform(train):
    transforms = []
    transforms.append(T.ToTensor())
    if train:
        transforms.append(T.RandomHorizontalFlip(0.5))
    return T.Compose(transforms)


def main():
    # train on the GPU or on the CPU, if a GPU is not available
    device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
	torch.manual_seed(1)
    # our dataset has two classes only - background and person
    num_classes = 2
    # use our dataset and defined transformations
    dataset = PennFudanDataset('PennFudanPed', get_transform(train=True))
    dataset_test = PennFudanDataset('PennFudanPed', get_transform(train=False))

    # split the dataset in train and test set
    indices = torch.randperm(len(dataset)).tolist()
    dataset = torch.utils.data.Subset(dataset, indices[:-5])
    dataset_test = torch.utils.data.Subset(dataset_test, indices[-5:])

    # define training and validation data loaders
    data_loader = torch.utils.data.DataLoader(
        dataset, batch_size=2, shuffle=True, num_workers=4,
        collate_fn=utils.collate_fn)

    data_loader_test = torch.utils.data.DataLoader(
        dataset_test, batch_size=1, shuffle=False, num_workers=4,
        collate_fn=utils.collate_fn)

    # get the model using our helper function
    model = get_model_instance_segmentation(num_classes)

    # move model to the right device
    model.to(device)

    # construct an optimizer
    params = [p for p in model.parameters() if p.requires_grad]
    optimizer = torch.optim.SGD(params, lr=0.005,
                                momentum=0.9, weight_decay=0.0005)
    # and a learning rate scheduler
    lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
                                                   step_size=3,
                                                   gamma=0.1)

    # let's train it for 10 epochs
    num_epochs = 10

    for epoch in range(num_epochs):
        # train for one epoch, printing every 10 iterations
        train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
        # update the learning rate
        lr_scheduler.step()
        # evaluate on the test dataset
        evaluate(model, data_loader_test, device=device)

    print("That's it!")


if __name__ == "__main__":
    main()

---

---

1. 查看CUDA版本，cmd命令nvcc --version ↩︎