1.创建如下目录
2.将train_annotations中的xml文件转成json文件
# coding=utf-8 import xml.etree.ElementTree as ET import os import json import collections coco = dict() coco['images'] = [] coco['type'] = 'instances' coco['annotations'] = [] coco['categories'] = [] #category_set = dict() image_set = set() image_id = 2019100001 # train:2018xxx; val:2019xxx; test:2020xxx category_item_id = 1 annotation_id = 1 category_set = ['people', 'bicycle', 'electric bicycle'] #这里填写好检测的类别 ''' def addCatItem(name): global category_item_id category_item = dict() category_item['supercategory'] = 'none' category_item_id += 1 category_item['id'] = category_item_id category_item['name'] = name coco['categories'].append(category_item) category_set[name] = category_item_id return category_item_id ''' def addCatItem(name): ''' 增加json格式中的categories部分 ''' global category_item_id category_item = collections.OrderedDict() category_item['supercategory'] = 'none' category_item['id'] = category_item_id category_item['name'] = name coco['categories'].append(category_item) category_item_id += 1 def addImgItem(file_name, size): global image_id if file_name is None: raise Exception('Could not find filename tag in xml file.') if size['width'] is None: raise Exception('Could not find width tag in xml file.') if size['height'] is None: raise Exception('Could not find height tag in xml file.') # image_item = dict() #按照一定的顺序，这里采用collections.OrderedDict() image_item = collections.OrderedDict() jpg_name = os.path.splitext(file_name)[0] + '.jpg' image_item['file_name'] = jpg_name image_item['width'] = size['width'] image_item['height'] = size['height'] image_item['id'] = image_id coco['images'].append(image_item) image_set.add(jpg_name) image_id = image_id + 1 return image_id def addAnnoItem(object_name, image_id, category_id, bbox): global annotation_id #annotation_item = dict() annotation_item = collections.OrderedDict() annotation_item['segmentation'] = [] seg = [] # bbox[] is x,y,w,h # left_top seg.append(bbox[0]) seg.append(bbox[1]) # left_bottom seg.append(bbox[0]) seg.append(bbox[1] + bbox[3]) # right_bottom seg.append(bbox[0] + bbox[2]) seg.append(bbox[1] + bbox[3]) # right_top seg.append(bbox[0] + bbox[2]) seg.append(bbox[1]) annotation_item['segmentation'].append(seg) annotation_item['area'] = bbox[2] * bbox[3] annotation_item['iscrowd'] = 0 annotation_item['image_id'] = image_id annotation_item['bbox'] = bbox annotation_item['category_id'] = category_id annotation_item['id'] = annotation_id annotation_item['ignore'] = 0 annotation_id += 1 coco['annotations'].append(annotation_item) def parseXmlFiles(xml_path): xmllist = os.listdir(xml_path) xmllist.sort() for f in xmllist: if not f.endswith('.xml'): continue bndbox = dict() size = dict() current_image_id = None current_category_id = None file_name = None size['width'] = None size['height'] = None size['depth'] = None xml_file = os.path.join(xml_path, f) print(xml_file) tree = ET.parse(xml_file) root = tree.getroot() # 抓根结点元素 if root.tag != 'annotation': # 根节点标签 raise Exception( 'pascal voc xml root element should be annotation, rather than {}'.format(root.tag)) # elem is <folder>, <filename>, <size>, <object> for elem in root: current_parent = elem.tag current_sub = None object_name = None # elem.tag, elem.attrib，elem.text if elem.tag == 'folder': continue if elem.tag == 'filename': file_name = elem.text if file_name in category_set: raise Exception('file_name duplicated') # add img item only after parse <size> tag elif current_image_id is None and file_name is not None and size['width'] is not None: if file_name not in image_set: current_image_id = addImgItem(file_name, size) # 图片信息 print('add image with {} and {}'.format(file_name, size)) else: raise Exception('duplicated image: {}'.format(file_name)) # subelem is <width>, <height>, <depth>, <name>, <bndbox> for subelem in elem: bndbox['xmin'] = None bndbox['xmax'] = None bndbox['ymin'] = None bndbox['ymax'] = None current_sub = subelem.tag if current_parent == 'object' and subelem.tag == 'name': object_name = subelem.text # if object_name not in category_set: # current_category_id = addCatItem(object_name) # else: #current_category_id = category_set[object_name] current_category_id = category_set.index( object_name) + 1 # index默认从0开始,但是json文件是从1开始，所以+1 elif current_parent == 'size': if size[subelem.tag] is not None: raise Exception('xml structure broken at size tag.') size[subelem.tag] = int(subelem.text) # option is <xmin>, <ymin>, <xmax>, <ymax>, when subelem is <bndbox> for option in subelem: if current_sub == 'bndbox': if bndbox[option.tag] is not None: raise Exception( 'xml structure corrupted at bndbox tag.') bndbox[option.tag] = int(option.text) # only after parse the <object> tag if bndbox['xmin'] is not None: if object_name is None: raise Exception('xml structure broken at bndbox tag') if current_image_id is None: raise Exception('xml structure broken at bndbox tag') if current_category_id is None: raise Exception('xml structure broken at bndbox tag') bbox = [] # x bbox.append(bndbox['xmin']) # y bbox.append(bndbox['ymin']) # w bbox.append(bndbox['xmax'] - bndbox['xmin']) # h bbox.append(bndbox['ymax'] - bndbox['ymin']) print( 'add annotation with {},{},{},{}'.format(object_name, current_image_id - 1, current_category_id, bbox)) addAnnoItem(object_name, current_image_id - 1, current_category_id, bbox) # categories部分 for categoryname in category_set: addCatItem(categoryname) if __name__ == '__main__': xml_path = 'dataset/train_anatations' json_file = 'VOC2007/anatations/voc_2007_train.json' # xml_path = 'dataset/test_anatation' # json_file = 'dataset/test.json' parseXmlFiles(xml_path) json.dump(coco, open(json_file, 'w'))
3.制作txt文件
# -*- coding: utf-8 -*- # @Author: zhiwei # @Date: 2019-01-31 09:38:58 # @Last Modified by: zhiwei # @Last Modified time: 2019-01-31 10:21:58 # import os import re fp1_path = "JPEGImages" f = open('VOCdevkit2020/VOC2020/ImageSets/Main/train.txt', 'w') s = "" i = 0 for filename in os.listdir(fp1_path): # bicycle_train.txt # if re.match('^bicycle.+', filename) != None: # s1 = os.path.splitext(os.path.basename(filename))[0] # s = s + s1 + ' ' + str(1) + " " # else: # s1 = os.path.splitext(os.path.basename(filename))[0] # s = s + s1 + ' ' + str(0) + " " # electric bicycle_train.txt # if "electric_bicycle" in filename: # s1 = os.path.splitext(os.path.basename(filename))[0] # s = s + s1 + ' ' + str(1) + " " # else: # s1 = os.path.splitext(os.path.basename(filename))[0] # s = s + s1 + ' ' + str(0) + " " # train.txt s1 = os.path.splitext(os.path.basename(filename))[0] s = s + s1 + " " print(s) f.write(s) print(len([name for name in os.listdir(fp1_path)]))
4.配置yarm文件
在Detectronconfigs12_2017_baselines目录下，复制文件retinanet_R-50-FPN_1x.yaml，到Detectronconfigsmy目录下重命名为retinanet_R-50-FPN_1x1.0.yaml
5.修改yarm文件 （retinanet_R-50-FPN_1x1.0.yaml）
MODEL:
TYPE: retinanet
CONV_BODY: FPN.add_fpn_ResNet50_conv5_body
NUM_CLASSES: 4
NUM_GPUS: 1
SOLVER:
WEIGHT_DECAY: 0.0001
LR_POLICY: steps_with_decay
BASE_LR: 0.001
GAMMA: 0.1
MAX_ITER: 1000
STEPS: [0, 600, 800]
FPN:
FPN_ON: True
MULTILEVEL_RPN: True
RPN_MAX_LEVEL: 7
RPN_MIN_LEVEL: 3
COARSEST_STRIDE: 128
EXTRA_CONV_LEVELS: True
RETINANET:
RETINANET_ON: True
NUM_CONVS: 4
ASPECT_RATIOS: (1.0, 2.0, 0.5)
SCALES_PER_OCTAVE: 3
ANCHOR_SCALE: 4
LOSS_GAMMA: 2.0
LOSS_ALPHA: 0.25
TRAIN:
WEIGHTS: /home/Desktop/test/trainMOdel/R-50.pkl
DATASETS: ('voc_2007_train',)
SCALES: (800,)
MAX_SIZE: 1333
RPN_STRADDLE_THRESH: -1 # default 0
TEST:
DATASETS: ('coco_2014_minival',)
SCALE: 800
MAX_SIZE: 1333
NMS: 0.5
RPN_PRE_NMS_TOP_N: 10000 # Per FPN level
RPN_POST_NMS_TOP_N: 2000
OUTPUT_DIR: .
我们这里只解释具体的几个参数的含义。
1）cfg，是配置文件，均存在于configs目录下。
在Model中：
MODEL:
TYPE: generalized_rcnn
CONV_BODY: FPN.add_fpn_ResNet50_conv5_body
NUM_CLASSES: 81
FASTER_RCNN: True
其中需要初学者注意的是NUM_CLASSES，对于customer的数据集，该值为 类别数+1，因此对于coco来说就是80+1
对于Mask网络，model部分还应加上
MASK_ON: True
设置GPU的数量：
NUM_GPUS: 1
SOLVER设置：
SOLVER:
WEIGHT_DECAY: 0.0001
LR_POLICY: steps_with_decay
BASE_LR: 0.0025
GAMMA: 0.1
MAX_ITER: 60000
STEPS: [0, 30000, 40000]
首先，对于训练次数而言，如果数据集不大，我们可以设置为几千次，如果像coco这类数据量较大，几万次还是有必要的。default的单GPU下为60000次。这里值得一提的是，对于多GPU下，MAX_ITER的次数与GPU的数量成反比。另一个需要说的参数是BASE_LR，初始的学习率对于网络训练很重要，太大会使网络不容易收敛到最小值，太小又会使网络收敛过慢。正如之前博客所言，通常取e-3到e-4比较安全。这里0.0025的取值，猜测是作者根据网络和数据多次实验所取的较优的配置。在多GPU下，BASE_LR的取值与GPU成正比。
FPN和FAST RCNN设置：
FPN:
FPN_ON: True
MULTILEVEL_ROIS: True
MULTILEVEL_RPN: True
FAST_RCNN:
ROI_BOX_HEAD: fast_rcnn_heads.add_roi_2mlp_head
ROI_XFORM_METHOD: RoIAlign
ROI_XFORM_RESOLUTION: 7
ROI_XFORM_SAMPLING_RATIO: 2
曾经尝试改动FPN，但是与此同时model也应该改动，正常情况下这两项在customer数据集上不需要改动。
Train 的设置：
TRAIN:
WEIGHTS: https://s3-us-west-2.amazonaws.com/detectron/ImageNetPretrained/MSRA/R-50.pkl
DATASETS: ('coco_2014_train',)
SCALES: (500,)
MAX_SIZE: 833
BATCH_SIZE_PER_IM: 256
RPN_PRE_NMS_TOP_N: 2000 # Per FPN level
这里要注意的是DATASETS中，自己的数据集需要在dataset_catalog.py中补充，其中数据集的生成方式github中也有详细说明
Test 的设置：
TEST:
DATASETS: ('coco_2014_minival',)
SCALE: 500
MAX_SIZE: 833
NMS: 0.5
RPN_PRE_NMS_TOP_N: 1000 # Per FPN level
RPN_POST_NMS_TOP_N: 1000
其中，需要注意的包括，SCALE和MAX_SIZE，从他人处得到经验，想在inference时提高准确率其中的一个办法就是换成高分辨率，那么这两个参数就是需要改动的了。
另外NMS的设置可以在非密集场景下减少重复出现的box
6.开始训练
cd到$Detectron/tools/目录下执行命令
python train_net.py --cfg ../configs/my2020/retinanet_R-50-FPN_1x.0.yaml OUTPUT_DIR /home/gaomh/Desktop/test/trainMOdel
--cfg：配置文件路径OUTPUT_DIR：训练的输出路径
接下来就是自行训练过程了~
7.遇到的问题
INFO loader.py: 126: Stopping enqueue thread INFO loader.py: 113: Stopping mini-batch loading thread INFO loader.py: 113: Stopping mini-batch loading thread INFO loader.py: 113: Stopping mini-batch loading thread INFO loader.py: 113: Stopping mini-batch loading thread Traceback (most recent call last): File "tools/train_net.py", line 132, in <module> main() File "tools/train_net.py", line 114, in main checkpoints = detectron.utils.train.train_model() File "/home/learner/github/detectron/detectron/utils/train.py", line 86, in train_model handle_critical_error(model, 'Loss is NaN')
修改BASE_LR的值从0.01改为0.001