本科生毕设实验记录

1. 手写数字识别

1.1. 训练配置设定

from easydict import EasyDict as edict

__C                           = edict()
# Consumers can get config by: from config import cfg

cfg                           = __C

# YOLO options
__C.YOLO                      = edict()

# Set the class name
__C.YOLO.CLASSES              = "./data/classes/yymnist.names"
__C.YOLO.ANCHORS              = "./data/anchors/basline_anchors.txt"
__C.YOLO.STRIDES              = [8, 16, 32]
__C.YOLO.ANCHOR_PER_SCALE     = 3
__C.YOLO.IOU_LOSS_THRESH      = 0.5

# Train options
__C.TRAIN                     = edict()

__C.TRAIN.ANNOT_PATH          = "./data/dataset/yymnist_train.txt"
__C.TRAIN.BATCH_SIZE          = 2
# __C.TRAIN.INPUT_SIZE            = [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
__C.TRAIN.INPUT_SIZE          = [416]
__C.TRAIN.DATA_AUG            = True
__C.TRAIN.LR_INIT             = 1e-3
__C.TRAIN.LR_END              = 1e-6
__C.TRAIN.WARMUP_EPOCHS       = 2
__C.TRAIN.EPOCHS              = 5

# TEST options
__C.TEST                      = edict()

__C.TEST.ANNOT_PATH           = "./data/dataset/yymnist_test.txt"
__C.TEST.BATCH_SIZE           = 2
__C.TEST.INPUT_SIZE           = 416
__C.TEST.DATA_AUG             = False
__C.TEST.DECTECTED_IMAGE_PATH = "./data/detection/"
__C.TEST.SCORE_THRESHOLD      = 0.3
__C.TEST.IOU_THRESHOLD        = 0.45

• batch_size改为2，太大会导致内存溢出
• 为了减少训练的总量和避免过拟合，将epochs设置为5
• 训练过程开启了数据增强

1.2. 数据集设置

• 训练集1000张，测试集200张，训练总数为1000/2*5=2500回合
• 训练时长2h

2. ；两类机器人识别

2.1. 训练配置设定

from easydict import EasyDict as edict

__C                           = edict()
# Consumers can get config by: from config import cfg

cfg                           = __C

# YOLO options
__C.YOLO                      = edict()

# Set the class name
# __C.YOLO.CLASSES              = "./data/classes/coco.names"
# __C.YOLO.CLASSES              = "./data/classes/yymnist.names"
# __C.YOLO.CLASSES              = "./data/classes/robot2.names"
__C.YOLO.CLASSES              = "./data/classes/robot.names"
__C.YOLO.ANCHORS              = "./data/anchors/basline_anchors.txt"
__C.YOLO.STRIDES              = [8, 16, 32]
__C.YOLO.ANCHOR_PER_SCALE     = 3
__C.YOLO.IOU_LOSS_THRESH      = 0.5

# Train options
__C.TRAIN                     = edict()
__C.TRAIN.ANNOT_PATH          = "./data/dataset/train.txt"
__C.TRAIN.BATCH_SIZE          = 2
# __C.TRAIN.INPUT_SIZE            = [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
__C.TRAIN.INPUT_SIZE          = [416]
__C.TRAIN.DATA_AUG            = True
__C.TRAIN.LR_INIT             = 1e-3
__C.TRAIN.LR_END              = 1e-6
__C.TRAIN.WARMUP_EPOCHS       = 2
__C.TRAIN.EPOCHS              = 2

# TEST options
__C.TEST                      = edict()

__C.TEST.ANNOT_PATH           = "./data/dataset/test.txt"
__C.TEST.BATCH_SIZE           = 2
__C.TEST.INPUT_SIZE           = 416
__C.TEST.DATA_AUG             = False
__C.TEST.DECTECTED_IMAGE_PATH = "./data/detection/"
__C.TEST.SCORE_THRESHOLD      = 0.3
__C.TEST.IOU_THRESHOLD        = 0.45

• batch_size改为2，太大会导致内存溢出
• 为了减少训练的总量和避免过拟合，将epochs设置为5

1.2. 数据集设置

• 每一类训练集1500张，一共3000张，测试集250张，一共500张，训练总数为3000/3*2=2000回合

3. 实验技巧

3.1. 需要修改的文件

# 配置文件的参数
D:\JupyterProjects\TensorFlow2.0-Examples-master\4-Object_Detection\YOLOV3\core\config.py
# 修改输入图片的尺寸
D:\JupyterProjects\TensorFlow2.0-Examples-master\4-Object_Detection\YOLOV3\test.py