def buildDataset(self, phase):
self.dataset = KittiDataset(root_dir=self.data_dir,
class_list=self.class_list,
boundary=self.boundary,
split=self.split)
if phase == 'val':
self.dataset.toggleVal()
import torch
import torch.nn as nn
import torch.optim as optim
import torch.utils.data as data
import numpy as np
import visdom
viz = visdom.Visdom(env='loss')
from complexYOLO import ComplexYOLO
from kitti import KittiDataset
from region_loss import RegionLoss
batch_size = 4
# dataset
dataset = KittiDataset(root='/home/ding/Documents/deeplearning/kitti',
set='train')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
model = ComplexYOLO()
model.cuda()
# define optimizer
optimizer = optim.SGD(model.parameters(),
lr=1e-5,
momentum=0.9,
weight_decay=0.0005)
# from loss2 import RegionLoss
from tensorboardX import SummaryWriter
from eval import draw_image_train
from config import cfg
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
writer = SummaryWriter()
batch_size = 8
data_dir = cfg.DATA_DIR
# dataset
dataset = KittiDataset(root=data_dir, set='train')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
# val data
val_set = KittiDataset(root=data_dir, set='val')
val_loader = data.DataLoader(val_set,
batch_size,
shuffle=True,
pin_memory=False)
# model = DataParallel(ComplexYOLO())
# model = ComplexYOLO()
model = torch.load('ComplexYOLO_epoch%s' % 130)
date_str = str(datetime.date.today())
logging_file = 'training_log_' + date_str + '.log'
logging.basicConfig(filename=logging_file,
level=logging.DEBUG,
format='%(asctime)s %(message)s')
batch_size = 12
# Remove old loggings in the tensorboard folder
ts_dir = './logs'
for ts_file in os.listdir(ts_dir):
ts_path = os.path.join(ts_dir, ts_file)
os.unlink(ts_path)
# dataset
dataset = KittiDataset(root='/home/yd2466/Complex-YOLO/data', set='train')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
model = ComplexYOLO()
model.cuda()
# define optimizer
optimizer = optim.SGD(model.parameters(),
lr=1e-5,
momentum=0.9,
weight_decay=0.0005)
# define the number of epochs
import torch import torch.nn as nn import torch.optim as optim import torch.utils.data as data import numpy as np from complexYOLO import ComplexYOLO from kitti import KittiDataset from region_loss import RegionLoss batch_size=12 # dataset dataset=KittiDataset(root='/home/ai/KITTI',set='train') data_loader = data.DataLoader(dataset, batch_size, shuffle=True) model = ComplexYOLO() model.cuda() # define optimizer optimizer = optim.Adam(model.parameters()) # define loss function region_loss = RegionLoss(num_classes=8, num_anchors=5) for epoch in range(200): for batch_idx, (rgb_map, target) in enumerate(data_loader):
import torch
import torch.nn as nn
import torch.optim as optim
import torch.utils.data as data
import numpy as np
from complexYOLO import ComplexYOLO
from kitti import KittiDataset
from region_loss import RegionLoss
batch_size = 12
# dataset
dataset = KittiDataset(root='/home/yuliu/KITTI', set='train')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
model = ComplexYOLO()
model.cuda()
# define optimizer
optimizer = optim.SGD(model.parameters(),
lr=1e-5,
momentum=0.9,
weight_decay=0.0005)
# define loss function
region_loss = RegionLoss(num_classes=8, num_anchors=5)
logging_file = 'training_log_' + date_str + '.log'
logging.basicConfig(filename=logging_file,
level=logging.DEBUG,
format='%(asctime)s %(message)s')
# Remove old loggings in the tensorboard folder
ts_dir = './logs'
for ts_file in os.listdir(ts_dir):
ts_path = os.path.join(ts_dir, ts_file)
os.unlink(ts_path)
dirname = os.path.dirname(__file__)
# Construct training dataset
train_data_path = os.path.join(dirname, 'data')
dataset = KittiDataset(root=train_data_path, set='train')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
# Initialize a model
model = ComplexYOLO()
model.cuda()
# Define optimizer
optimizer = optim.SGD(model.parameters(),
lr=lr,
momentum=momentum,
weight_decay=weight_decay)
import torch
import torch.nn as nn
import torch.optim as optim
import torch.utils.data as data
import numpy as np
from complexYOLO import ComplexYOLO
from kitti import KittiDataset
from region_loss import RegionLoss
batch_size = 12
# dataset
dataset = KittiDataset(root='./', set='train')
data_loader = data.DataLoader(dataset, batch_size, shuffle=True)
model = ComplexYOLO()
model.cuda()
# define optimizer
optimizer = optim.Adam(model.parameters(), lr=0.001)
# define loss function
region_loss = RegionLoss(num_classes=8, num_anchors=5)
for epoch in range(20):
for batch_idx, (rgb_map, target) in enumerate(data_loader):
optimizer.zero_grad()
rgb_map = rgb_map.view(rgb_map.data.size(0), rgb_map.data.size(3),
"""
@description:这是测试输出中间层特征得例子
"""
import torch
from kitti import KittiDataset
import torch.utils.data as data
import scipy.misc as misc
batch_size = 12
model = torch.load('ComplexYOLO_epoch%s' % 130)
model.cuda()
print(model)
# dataset
dataset = KittiDataset(root='/home/yxk/data/Kitti/object', set='train')
data_loader = data.DataLoader(dataset,
batch_size,
num_workers=6,
shuffle=True,
pin_memory=False)
for batch_idx, (rgb_map, target) in enumerate(data_loader):
rgb_map = rgb_map.view(rgb_map.data.size(0), rgb_map.data.size(3),
rgb_map.data.size(1), rgb_map.data.size(2))
x = rgb_map.float().cuda()
for name, module in model._modules.items():
x = module(x)
print(x.shape)
model = torch.load('ComplexYOLO_epoch%s' % 70)
import torch.nn as nn
import torch.optim as optim
import torch.utils.data as data
import numpy as np
from complexYOLO import ComplexYOLO
from kitti import KittiDataset
from region_loss import RegionLoss
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
batch_size = 12
# dataset
dataset = KittiDataset(root='/home/Documents/training/training/', set='train')
data_loader = data.DataLoader(dataset, batch_size, shuffle=True)
model = ComplexYOLO()
model.cuda()
# define optimizer
optimizer = optim.Adam(model.parameters())
# define loss function
region_loss = RegionLoss(num_classes=8, num_anchors=5)
for epoch in range(400):
for batch_idx, (rgb_map, target) in enumerate(data_loader):
optimizer.zero_grad()
class Solver():
def __init__(self, args):
self.data_dir = args.data_dir
self.log = args.do_log
self.split = args.split
self.epochs = args.epochs
self.batch_size = args.batch_size
self.lr = args.lr
self.momentum = args.momentum
self.weight_decay = args.weight_decay
with open('config.json', 'r') as f:
config = json.load(f)
self.boundary = config["boundary"]
self.class_list = config["class_list"]
self.anchors = config["anchors"]
self.num_anchors = len(self.anchors) // 2
self.num_classes = len(self.class_list) + 1
self.save_dir = args.save_dir
self.device = torch.device(
'cuda' if torch.cuda.is_available() else 'cpu')
self.buildModel()
def buildModel(self):
self.model = ComplexYOLO()
self.model.to(self.device)
def buildDataset(self, phase):
self.dataset = KittiDataset(root_dir=self.data_dir,
class_list=self.class_list,
boundary=self.boundary,
split=self.split)
if phase == 'val':
self.dataset.toggleVal()
def buildLogger(self):
pass
def train(self):
self.buildDataset('train')
opt = optim.SGD(self.model.parameters(), lr=self.lr)
region_loss = RegionLoss(num_classes=self.num_classes,
num_anchors=self.num_anchors)
region_loss = region_loss.to(self.device)
# Dataloader
data_loader = DataLoader(self.dataset,
self.batch_size,
shuffle=True,
pin_memory=False)
log_step = 10
best_loss = 10000
self.model.train()
for epoch in range(self.epochs):
running_loss = 0
running_nCorrect = 0
running_nGT = 0
for idx, data in enumerate(data_loader):
# pdb.set_trace()
rgb_map = data[0]
target = data[1]
rgb_map = rgb_map.float().to(self.device)
output = self.model(rgb_map)
loss, nCorrect, nGT = region_loss(output, target)
loss.backward()
running_loss += loss.item()
running_nCorrect += nGT
running_nCorrect += nCorrect
if idx % log_step == 0:
mean_loss = running_loss / log_step
print("Epoch: {}, Loss: {m:=5.4f}".format(epoch,
m=mean_loss))
print("nCorrect = {m:=5.4f}, nGT = {p:=5.4f}".format(
m=running_nCorrect / log_step,
p=running_nGT / log_step))
if mean_loss < best_loss:
best_loss = mean_loss
path = os.path.join(
self.save_dir,
'ep-{}-{m:=5.4f}.pth'.format(epoch, m=mean_loss))
torch.save(self.model.state_dict(), path)
print("Saved model for {} epoch\n".format(epoch))
running_loss = 0
running_nCorrect = 0
running_nGT = 0
opt.step()
def test(self):
pass
from config import cfg from threading import Thread from kitti import KittiDataset from visualization import vis_points from load_points import load_kitti_velodyne, get_filtered_lidar, kitti_velo_to_cam dataset = KittiDataset(cfg) index = dataset._load_index() print(len(index), type(index)) print(index) print(dataset.label_path_at(5)) print(dataset.lidar_path_at(5)) print(dataset.image_path_at(5)) print(dataset.calib_path_at(5)) origin = load_kitti_velodyne(dataset.lidar_path_at(5)) crop = get_filtered_lidar(origin) # vis_mayavi(origin) xyzi = kitti_velo_to_cam(origin, dataset.calib_path_at(5)) vis_points(xyzi)
format='%(asctime)s %(message)s')
# Remove old loggings in the tensorboard folder
# ts_dir = './logs'
# for ts_file in os.listdir(ts_dir):
# ts_path = os.path.join(ts_dir, ts_file)
# os.unlink(ts_path)
dirname = os.path.dirname(__file__)
# Construct training dataset
with open('config.json', 'r') as f:
config = json.load(f)
train_data_path = os.path.join(dirname, 'data')
dataset = KittiDataset(root_dir=args.data_dir,
class_list=config["class_list"],
boundary=config["boundary"],
set='lidar')
data_loader = data.DataLoader(dataset,
batch_size,
shuffle=True,
pin_memory=False)
# Initialize a model
model = ComplexYOLO()
model.cuda()
# Define optimizer
optimizer = optim.SGD(model.parameters(),
lr=lr,
momentum=momentum,
weight_decay=weight_decay)