基础配置

1. 首先你需要在win10上下载Git（用于我们在github上面下载源码）和MinGW（方便我们在win10上也能用linux的make操作命令）。
2. 接着你要下载cuda9.0和cudnn7.1来绑定你的windows的Nvidia
3. 接着你需要在win10上面安装anaconda3（切记，python用的是3.6+，目前的tesorflow-gpu只能匹配这个）
4. 然后在现有的 环境下(或者配置新环境），按照顺序依次用conda install cudatoolkit==9.0->conda install cudnn==7.1.4->conda install tensorflow-gpu==1.9.0->conda install keras-gpu==2.2.4
5. 然后记住要装一些其他需要的包，比如（matplotlib、openCV、cython、scikit-image等等），总之就是按照提示conda install或者pip install

前期需要下载的内容

1. 首先大家用在github上把mask r-cnn的项目下载到本地；
2. 然后要把cocoapi下载到本地；
3. 接着是把预权重mask_rcnn_coco.h5放到mask r-cnn文件下面的新建文件夹logs下面；
4. 最后是打开anaconda prompt，cd到cocoapi/PythonAPI的目录，一次输入make和make -j8；
5. 把PythonAPI下面更新的pycocotools放到mask r-cnn文件的samples/coco/里；

先附上mask-rcnn的官方测试代码：

import os
import sys
import random
import math
import numpy as np
import skimage.io
import matplotlib
import matplotlib.pyplot as plt
import cv2
import time

ROOT_DIR = os.path.abspath(\"G:\\Mask_RCNN\")

sys.path.append(ROOT_DIR)
from mrcnn import utils
import mrcnn.model as modellib
from mrcnn import visualize

sys.path.append(os.path.join(ROOT_DIR, \"samples/coco/\"))
import coco

MODEL_DIR = os.path.join(ROOT_DIR, \"logs\")

COCO_MODEL_PATH = os.path.join(MODEL_DIR ,\"mask_rcnn_coco.h5\")

if not os.path.exists(COCO_MODEL_PATH):
    utils.download_trained_weights(COCO_MODEL_PATH)
    print(\"cuiwei***********************\")

IMAGE_DIR = os.path.join(ROOT_DIR, \"images\")

class InferenceConfig(coco.CocoConfig):
    GPU_COUNT = 1
    IMAGES_PER_GPU = 1
    
config = InferenceConfig()
config.display()

model = modellib.MaskRCNN(mode=\"inference\", model_dir=MODEL_DIR, config=config)

#model.load_weights(COCO_MODEL_PATH, by_name=True)
model.load_weights(COCO_MODEL_PATH, by_name=True, exclude=[ \"mrcnn_class_logits\", \"mrcnn_bbox_fc\"])
# =============================================================================
# class_names = [\'BG\', \'person\', \'bicycle\', \'car\', \'motorcycle\', \'airplane\',
#                \'bus\', \'train\', \'truck\', \'boat\', \'traffic light\',
#                \'fire hydrant\', \'stop sign\', \'parking meter\', \'bench\', \'bird\',
#                \'cat\', \'dog\', \'horse\', \'sheep\', \'cow\', \'elephant\', \'bear\',
#                \'zebra\', \'giraffe\', \'backpack\', \'umbrella\', \'handbag\', \'tie\',
#                \'suitcase\', \'frisbee\', \'skis\', \'snowboard\', \'sports ball\',
#                \'kite\', \' ball bat\', \' ball glove\', \'skateboard\',
#                \'surfboard\', \'tennis racket\', \'bottle\', \'wine glass\', \'cup\',
#                \'fork\', \'knife\', \'spoon\', \'bowl\', \'banana\', \'apple\',
#                \'sandwich\', \'orange\', \'broccoli\', \'carrot\', \'hot dog\', \'pizza\',
#                \'donut\', \'cake\', \'chair\', \'couch\', \'potted plant\', \'bed\',
#                \'dining table\', \'toilet\', \'tv\', \'laptop\', \'mouse\', \'remote\',
#                \'keyboard\', \'cell phone\', \'microwave\', \'oven\', \'toaster\',
#                \'sink\', \'refrigerator\', \'book\', \'clock\', \'vase\', \'scissors\',
#                \'teddy bear\', \'hair drier\', \'toothbrush\']
# =============================================================================
class_names = [\'BG\', \'person\']
#file_names=next(os.walk(IMAGE_DIR))[2]
#image=skimage.io.imread(os.path.join(IMAGE_DIR,random.choice(file_names)))
image= cv2.imread(\"G:\\\\Mask_RCNN\\\\images\\\\25691390_f9944f61b5_z.jpg\")

#cap = cv2.VideoCapture(0)
i=0
while(i<1):
    i+=1
#    ret,   = cap.read()
    
    start =time.clock()
#    results = model.detect([ ], verbose=1)
    results=model.detect([image],verbose=1)
    
    r = results[0]
    
#    visualize.display_instances( , r[\'rois\'], r[\'masks\'], r[\'class_ids\'], class_names, r[\'scores\'])
    visualize.display_instances(image,r[\'rois\'],r[\'masks\'],r[\'class_ids\'],class_names,r[\'scores\'])
    end = time.clock()
    print(end-start)
    if cv2.waitKey(1) & 0xFF == ord(\'q\'):
        break
    
#cap.release()
cv2.destroyAllWindows()

在测试代码中会踩到的坑

Traceback (most recent call last):

  File \"<ipython-input-1-fdee81fb82fb>\", line 1, in <module>
    runfile(\'G:/labelme/test.py\', wdir=\'G:/labelme\')

  File \"C:\\Users\\34905\\Anaconda3\\envs\\cv2\\lib\\site-packages\\spyder_kernels\\customize\\spydercustomize.py\", line 704, in runfile
    execfile(filename, namespace)

  File \"C:\\Users\\34905\\Anaconda3\\envs\\cv2\\lib\\site-packages\\spyder_kernels\\customize\\spydercustomize.py\", line 108, in execfile
    exec(compile(f.read(), filename, \'exec\'), namespace)

  File \"G:/labelme/test.py\", line 48, in <module>
    model.load_weights(COCO_MODEL_PATH, by_name=True)

  File \"G:\\Mask_RCNN\\mrcnn\\model.py\", line 2131, in load_weights
    saving.load_weights_from_hdf5_group_by_name(f,  s)

  File \"C:\\Users\\34905\\Anaconda3\\envs\\cv2\\lib\\site-packages\\keras\\engine\\saving.py\", line 1149, in load_weights_from_hdf5_group_by_name
    str(weight_values[i].shape) + \'.\')

ValueError:   #389 (named \"mrcnn_bbox_fc\"), weight <tf.Variable \'mrcnn_bbox_fc/kernel:0\' shape=(1024, 8) dtype=float32_ref> has shape (1024, 8), but the saved weight has shape (1024, 324).

如果你不想测试coco里默认的81类，只想测试2类，那一定记住要把model.load_weights(COCO_MODEL_PATH, by_name=True)改为model.load_weights(COCO_MODEL_PATH, by_name=True, exclude=[\"mrcnn_class_logits\", \"mrcnn_bbox_fc\",\"mrcnn_bbox\", \"mrcnn_mask\"])

附上训练代码

import os
import sys
import random
import math
import re
import time
import numpy as np
import cv2
import matplotlib
import matplotlib.pyplot as plt
import tensorflow as tf

#ROOT_DIR = os.getcwd()
ROOT_DIR = os.path.abspath(\"G:\\Mask_RCNN\")
sys.path.append(ROOT_DIR)

from mrcnn.config import Config
#import utils
from mrcnn import model as modellib,utils
from mrcnn import visualize
import yaml
from mrcnn.model import log
from PIL import Image


MODEL_DIR = os.path.join(ROOT_DIR, \"logs\")

iter_num=0

COCO_MODEL_PATH = os.path.join(MODEL_DIR, \"mask_rcnn_coco.h5\")

if not os.path.exists(COCO_MODEL_PATH):
    utils.download_trained_weights(COCO_MODEL_PATH)
    
class ShapesConfig(Config):
     NAME = \"pods\"
    
     GPU_COUNT = 1
     
     IMAGES_PER_GPU = 2
    
     NUM_CLASSES = 1 + 1
     
     IMAGE_MIN_DIM = 320
     IMAGE_MAX_DIM = 384
     
#     RPN_ANCHOR_SCALES = (8 * 6, 16 * 6, 32 * 6, 64 * 6, 128 * 6)
     RPN_ANCHOR_SCALES = (32, 64, 128, 256, 512)
     
     RPN_ANCHOR_RATIOS = [0.5, 1, 2]
     
     RPN_ANCHOR_STRIDE = 1
     RPN_NMS_THRESHOLD = 0.7
     RPN_TRAIN_ANCHORS_PER_IMAGE = 256
     
     TRAIN_ROIS_PER_IMAGE = 100
     
     STEPS_PER_EPOCH = 28
     
     MAX_GT_INSTANCES = 100
     
     VALIDATION_STEPS = 5
config = ShapesConfig()
config.display()

class DrugDataset(utils.Dataset):
    def get_obj_index(self, image):
        n = np.max(image)
        return n
    
    def from_yaml_get_class(self, image_id):
        info = self.image_info[image_id]
        with open(info[\'yaml_path\']) as f:
            temp = yaml.load(f.read())
            labels = temp[\'label_names\']
            del labels[0]
        return labels
    
    def draw_mask(self, num_obj, mask, image,image_id):
        info = self.image_info[image_id]
        for index in range(num_obj):
            for i in range(info[\'width\']):
                for j in range(info[\'height\']):
                    at_pixel = image.getpixel((i, j))
                    if at_pixel == index + 1:
                        mask[j, i, index] = 1
        return mask
    
    def load_shapes(self, count, img_folder, mask_folder, imglist, dataset_root_path):
        self.add_class(\"shapes\", 1, \"pod\")
        for i in range(count):
            filestr = imglist[i].split(\".\")[0]
            mask_path = mask_folder + \"/\" + filestr + \".png\"
            yaml_path = dataset_root_path + \"labelme_json/\" + filestr + \"_json/info.yaml\"
            print(dataset_root_path + \"labelme_json/\" + filestr + \"_json/img.png\")
            cv_img = cv2.imread(dataset_root_path + \"labelme_json/\" + filestr + \"_json/img.png\")
            
            self.add_image(\"shapes\", image_id=i, path=img_folder + \"/\" + imglist[i],
                           width=cv_img.shape[1], height=cv_img.shape[0], mask_path=mask_path, yaml_path=yaml_path)

    def load_mask(self, image_id):
        global iter_num
        print(\"image_id\",image_id)
        info = self.image_info[image_id]
        count = 1  # number of  
        img = Image.open(info[\'mask_path\'])
        num_obj = self.get_obj_index(img)
        mask = np.zeros([info[\'height\'], info[\'width\'], num_obj], dtype=np.uint8)
        mask = self.draw_mask(num_obj, mask, img,image_id)
        occlusion = np.logical_not(mask[:, :, -1]).astype(np.uint8)
        for i in range(count - 2, -1, -1):
            mask[:, :, i] = mask[:, :, i] * occlusion
            
            occlusion = np.logical_and(occlusion, np.logical_not(mask[:, :, i]))
        labels = []
        labels = self.from_yaml_get_class(image_id)
        labels_form = []
        for i in range(len(labels)):
            if labels[i].find(\"pod\") != -1:
                # print \"box\"
                labels_form.append(\"pod\")
        class_ids = np.array([self.class_names.index(s) for s in labels_form])
        return mask, class_ids.astype(np.int32)

def get_ax(rows=1, cols=1, size=8):
    _, ax = plt.subplots(rows, cols, figsize=(size * cols, size * rows))
    return ax

#dataset_root_path=\"train_data/\"
dataset_root_path = os.path.join(ROOT_DIR, \"train_data/\")
img_folder = dataset_root_path + \"pic\"
mask_folder = dataset_root_path + \"cv2_mask\"
imglist = os.listdir(img_folder)
count = len(imglist)

dataset_train = DrugDataset()
dataset_train.load_shapes(count, img_folder, mask_folder, imglist,dataset_root_path)
dataset_train.prepare()

dataset_val = DrugDataset()
dataset_val.load_shapes(56, img_folder, mask_folder, imglist,dataset_root_path)
dataset_val.prepare()

model = modellib.MaskRCNN(mode=\"training\", config=config,model_dir=MODEL_DIR)

init_with = \"coco\"

if init_with == \"imagenet\":
    model.load_weights(model.get_imagenet_weights(), by_name=True)
elif init_with == \"coco\":
    model.load_weights(COCO_MODEL_PATH, by_name=True,exclude=[\"mrcnn_class_logits\", \"mrcnn_bbox_fc\",\"mrcnn_bbox\", \"mrcnn_mask\"])
elif init_with == \"last\":
    model.load_weights(model.find_last()[1], by_name=True)
    
#model.train(dataset_train, dataset_val,learning_rate=config.LEARNING_RATE,epochs=1, s=\'heads\')
model.train(dataset_train, dataset_val,learning_rate=config.LEARNING_RATE,epochs=20, s=\'heads\')

model.train(dataset_train, dataset_val,learning_rate=config.LEARNING_RATE / 10,epochs=20, s=\"all\")
#model.train(dataset_train, dataset_val,learning_rate=config.LEARNING_RATE / 10,epochs=1, s=\"all\")

训练代码中遇到的坑

Traceback (most recent call last):

  File \"<ipython-input-10-4a900268c3f6>\", line 1, in <module>
    runfile(\'D:/Mask R-CNN/train.py\', wdir=\'D:/Mask R-CNN\')

  File \"D:\\Anaconda3\\Anaconda3-5.3.0\\envs\\cv2\\lib\\site-packages\\spyder_kernels\\customize\\spydercustomize.py\", line 704, in runfile
    execfile(filename, namespace)

  File \"D:\\Anaconda3\\Anaconda3-5.3.0\\envs\\cv2\\lib\\site-packages\\spyder_kernels\\customize\\spydercustomize.py\", line 108, in execfile
    exec(compile(f.read(), filename, \'exec\'), namespace)

  File \"D:/Mask R-CNN/train.py\", line 158, in <module>
    model.load_weights(COCO_MODEL_PATH, by_name=True,exclude=[\"mrcnn_class_logits\", \"mrcnn_bbox_fc\",\"mrcnn_bbox\", \"mrcnn_mask\"])

  File \"D:\\Mask R-CNN\\mrcnn\\model.py\", line 2131, in load_weights
    saving.load_weights_from_hdf5_group_by_name(f,  s)

  File \"D:\\Anaconda3\\Anaconda3-5.3.0\\envs\\cv2\\lib\\site-packages\\keras\\engine\\saving.py\", line 1104, in load_weights_from_hdf5_group_by_name
    g = f[name]

  File \"h5py\\_ s.pyx\", line 54, in h5py._ s.with_phil.wrapper

  File \"h5py\\_ s.pyx\", line 55, in h5py._ s.with_phil.wrapper

  File \"D:\\Anaconda3\\Anaconda3-5.3.0\\envs\\cv2\\lib\\site-packages\\h5py\\_hl\\group.py\", line 177, in __getitem__
    oid = h5o.open(self.id, self._e(name), lapl=self._lapl)

  File \"h5py\\_ s.pyx\", line 54, in h5py._ s.with_phil.wrapper

  File \"h5py\\_ s.pyx\", line 55, in h5py._ s.with_phil.wrapper

  File \"h5py\\h5o.pyx\", line 190, in h5py.h5o.open

KeyError: \'Unable to open   (wrong B-tree signature)\'

这个bug是我目前遇到最难察觉问题的bug，而且网上关于这个bug的解决方法很少，经过我仔细的排查，我发现是我的mask_rcnn_coco.h5文件有问题。我重新下载后，就解决了这个bug，建议大家在重新搭建平台的时候，不要用上一个平台已经用过的预权重 mask_rcnn_coco.h5，会发生很多不可知的错误。最好还是重新去网上下载一个。

测试自己的训练代码

import os
import sys
import random
import math
import numpy as np
import skimage.io
import matplotlib
import matplotlib.pyplot as plt
import cv2
import time

ROOT_DIR = os.path.abspath(\"G:\\Mask_RCNN\")

sys.path.append(ROOT_DIR)

from mrcnn.config import Config
from datetime import datetime
from mrcnn import utils
import mrcnn.model as modellib
from mrcnn import visualize

sys.path.append(os.path.join(ROOT_DIR, \"samples/coco/\"))
from samples.coco import coco

MODEL_DIR = os.path.join(ROOT_DIR, \"logs/shapes20181122T2257\")
#MODEL_DIR = os.path.join(ROOT_DIR, \"logs\")

COCO_MODEL_PATH = os.path.join(MODEL_DIR, \"mask_rcnn_shapes_0011.h5\")
#COCO_MODEL_PATH = os.path.join(MODEL_DIR, \"mask_rcnn_coco.h5\")

IMAGE_DIR = os.path.join(ROOT_DIR, \"train_data/pic\")
#IMAGE_DIR = os.path.join(ROOT_DIR, \"images\")

class ShapesConfig(Config):
    NAME = \"shapes\"
    
    GPU_COUNT = 1
    IMAGES_PER_GPU = 1
    
    NUM_CLASSES = 1 + 1
    
    IMAGE_MIN_DIM = 320
    IMAGE_MAX_DIM = 384
    
    RPN_ANCHOR_SCALES = (8 * 6, 16 * 6, 32 * 6, 64 * 6, 128 * 6)
    
    TRAIN_ROIS_PER_IMAGE =100
    
    STEPS_PER_EPOCH = 10
    
    VALIDATION_STEPS = 10
    
class InferenceConfig(ShapesConfig):
    
    GPU_COUNT = 1
    IMAGES_PER_GPU = 1
    
config = InferenceConfig()

model = modellib.MaskRCNN(mode=\"inference\", model_dir=MODEL_DIR, config=config)

model = modellib.MaskRCNN(mode=\"inference\", model_dir=MODEL_DIR, config=config)

model.load_weights(COCO_MODEL_PATH, by_name=True)

class_names = [\'BG\', \'pod\']
i=1
while(i<2):
    i=i+1
    file_names = next(os.walk(IMAGE_DIR))[2]
#    image = skimage.io.imread(os.path.join(IMAGE_DIR, random.choice(file_names)))
    image = skimage.io.imread(os.path.join(IMAGE_DIR, \"1001-01b.jpg\"))
    a=datetime.now()

    results = model.detect([image], verbose=1)
    b=datetime.now()

    print(\"time:\",(b-a).seconds)
    r = results[0]
    visualize.display_instances(image, r[\'rois\'], r[\'masks\'], r[\'class_ids\'], class_names, r[\'scores\'])

同样的在Linux的ubutu16.04上测试我的代码的时候遇到了这个问题。

遇到这个问题大家不要慌，这是因为大家设置了多线程，但是线程不同步造成的。大家耐心等一会，就会出现loss了。同时如果大家不希望出现多线程，大家可以改为单线程。更改方法如下：在Mask RCNN\\mrcnn\\model.py中

 self.keras_model.fit_generator(
            train_generator,
            initial_epoch=self.epoch,
            epochs=epochs,
            steps_per_epoch=self.config.STEPS_PER_EPOCH,
            callbacks=callbacks,
            validation_data=val_generator,
            validation_steps=self.config.VALIDATION_STEPS,
            max_queue_size=100,
            workers=workers,
            use_multiprocessing=True,
#            use_multiprocessing=False,
        )

大家如果只是想要把多线程改为单线程，就要把use_multiprocessing=False,同时要让workers=1。因为单线程的情况下让workers大于1会报错！

有什么其他问题欢迎大家在下方留言私信我！