def recog(fname_or_list, train=None):
"""
识别 demo，输入一个或多个视频文件路径，返回识别结果
"""
if train is None:
train = trainer(is_recog=True)
train.saver.restore(train.sess, train.model_fname)
if not isinstance(fname_or_list, list):
fname_or_list = [fname_or_list]
with timer('识别视频') as t:
probs, labels = train.recog(fname_or_list)
delta = t.duration()
is_all_test_set = False
if fname_or_list == train.batch_generator.test_fname_list:
is_all_test_set = True
top_accs = [0 for _ in range(5)]
for i in range(len(probs)):
print('--------------------------------------------------')
fname = fname_or_list[i]
prob = probs[i]
label = labels[i]
cls_name = train.batch_generator.cls_list[label]
str_format = '%s 的预测标签为：%d，预测类别为：%s' % (fname, label, cls_name)
print(str_format)
prob = prob.tolist()
if is_all_test_set:
true_label = train.batch_generator.test_fname_to_label[fname]
print('Top-5 预测概率值分别为：')
top_5_index = np.argsort(prob)[::-1][:5]
for i in range(0, len(top_5_index), 2):
index = top_5_index[i]
if is_all_test_set:
if true_label in top_5_index[:i+1]:
top_accs[i] += 1
str_format = '标签：%d - 类别名：%s - 概率值：%f'
cls_name = train.batch_generator.cls_list[index]
p = prob[index]
print(str_format % (index, cls_name, p))
print('--------------------------------------------------')
print('以上为所有输入视频的预测标签、类别名和预测概率值')
print('总视频数为：%d' % len(fname_or_list))
print('视频总用时：%.3f' % delta)
speed = delta / len(fname_or_list)
print('识别平均速度为：%.3f sec/视频' % speed)
if is_all_test_set:
print('++++++++++++++++++++++++++++++++++++++++++++++++++')
print('当前使用所有测试集作为演示，可计算准确率，准确率如下：')
for i in range(0, len(top_accs), 2):
print('Top-%d准确率为：%.4f' % (i+1, top_accs[i] / len(fname_or_list)))
train.close()
import sys
import bmp
import prep
import tools
import solver
import postpro
import deformed
import gallery
import version
import prob
import copy
import plast
import gc
#start timer
TIME = tools.timer()
#welcome message
print("")
print("GRoT> ver. " + version.get() + ", [Graficzny Rozwiązywacz Tarcz]")
print("................................................\n")
#read optional command line arguments for input and output path folder
PROJECT_PATH = False #project will contain input bmp image and input.txt with parameters
RESULT_PATH = False #result will contain output of grot
args = sys.argv
if len(args) > 1:
for i in range(len(args)):
if args[i] == "-i":
PROJECT_PATH = args[i + 1]
if __name__ == '__main__':
# verify GPU is available
if not cuda.is_available():
raise CudaSupportError("No CUDA-compatible device found")
# initialize coordinate grid (in degrees)
ha0 = 30.
dec0 = 45.
dhacosdec = np.linspace(-.5, .5, 128, dtype=np.float32)
ddec = np.linspace(-.5, .5, 128, dtype=np.float32)
dHACOSDEC, dDEC = np.meshgrid(dhacosdec, ddec)
# run beamformer
with timer('GPU'):
bf_gpu = BeamformerGPU(ha0, dec0, dHACOSDEC, dDEC)
# bf_gpu.tabs shape is (ntab, nfreq, ndec, nha)
# select one dec
bf_gpu.tabs = bf_gpu.tabs[:, :, 63]
# plot a few tabs
tabs_to_plot = [0, 3, 6, 9]
fig, axes = plt.subplots(nrows=2,
ncols=2,
figsize=(8, 8),
sharex=True,
sharey=True)
axes = axes.flatten()
X, Y = np.meshgrid(dhacosdec, bf_gpu.freqs)
from tools import timer
from datetime import datetime
# print(str(sys.argv))
# try:
# s = sys.argv[1]
# except:
# pass
# print(s)
# exit()
col_news = db["news"]
col_rss = db["rss"]
col_rss_log = db["rss_log"]
t = timer()
# col_rss = db['rss']
# for item in rss_links:
# item['handy'] = 1
# item['active'] = 1
# col_rss.insert(item)
# exit()
from lxml import etree
parser = etree.HTMLParser()
tree = etree.parse(
"http://www.dailymail.co.uk/sport/football/article-3437261/Real-Madrid-just-11-fans-attend-win-Granada-Spanish-averse-travelling-away-games.html?ITO=1490&ns_mchannel=rss&ns_campaign=1490",
parser,
target = target[:10000]
test = test[:10000]
# Actual Run Code
lable_group = LabelEncoder()
Y = lable_group.fit_transform(target) # 变成一个列表
classes_ = lable_group.classes_
print('Y\n', Y)
print('classes_\n', classes_)
# enter the number of folds from xgb.cv
ntest = test.shape[0]
folds = 5
early_stopping = 50
oof_test = np.zeros((ntest, 6))
start_time = timer(None)
# Load data set and target values
# test = test.loc[:, ~test.columns.duplicated()]
# print('test\n', test)
d_test = xgb.DMatrix(test.values)
# In[ ]:
# set up KFold that matches xgb.cv number of folds
kf = StratifiedKFold(n_splits=folds, random_state=0)
# Start the CV
acc = []
for i, (train_index, test_index) in enumerate(kf.split(train, Y)):
from tools import timer
from datetime import datetime
# print(str(sys.argv))
# try:
# s = sys.argv[1]
# except:
# pass
# print(s)
# exit()
col_news = db['news']
col_rss = db['rss']
col_rss_log = db['rss_log']
t = timer()
# col_rss = db['rss']
# for item in rss_links:
# item['handy'] = 1
# item['active'] = 1
# col_rss.insert(item)
# exit()
from lxml import etree
parser = etree.HTMLParser()
tree = etree.parse("http://www.dailymail.co.uk/sport/football/article-3437261/Real-Madrid-just-11-fans-attend-win-Granada-Spanish-averse-travelling-away-games.html?ITO=1490&ns_mchannel=rss&ns_campaign=1490", parser)
print(tree)
# reference frequency for dedispersion, use highest frequency
ref_freq = freqs.max()
# construct dm range to search around known value
dm_offsets = np.arange(-16, 16, .1, dtype=np.float32)
ndm = len(dm_offsets)
print("CUDA info:")
cuda.detect()
print("DMs: {}".format(ndm))
print("Freqs: {}".format(nfreq))
print("Times: {}".format(ntime))
# serial CPU. CPU is assumed to be right
with timer('CPU'):
truth = dedisp_cpu(data, dm_offsets, freqs)
# Parallelized CPU.
with timer('CPU parallel'):
cpu_par = dedisp_cpu_par(data, dm_offsets, freqs)
assert np.allclose(cpu_par, truth, atol=1E-6)
# GPU
gpu = cuda.get_current_device()
nthread = (gpu.WARP_SIZE, gpu.WARP_SIZE)
nx = math.ceil(ndm / nthread[0])
ny = math.ceil(ntime / nthread[1])
nblock = (nx, ny)
output_gpu = np.empty((ndm, ntime), dtype=np.float32)
top_accs[i] += 1
str_format = '标签：%d - 类别名：%s - 概率值：%f'
cls_name = train.batch_generator.cls_list[index]
p = prob[index]
print(str_format % (index, cls_name, p))
print('--------------------------------------------------')
print('以上为所有输入视频的预测标签、类别名和预测概率值')
print('总视频数为：%d' % len(fname_or_list))
print('视频总用时：%.3f' % delta)
speed = delta / len(fname_or_list)
print('识别平均速度为：%.3f sec/视频' % speed)
if is_all_test_set:
print('++++++++++++++++++++++++++++++++++++++++++++++++++')
print('当前使用所有测试集作为演示，可计算准确率，准确率如下：')
for i in range(0, len(top_accs), 2):
print('Top-%d准确率为：%.4f' % (i+1, top_accs[i] / len(fname_or_list)))
train.close()
if __name__ == '__main__':
fname_or_list = sys.argv[1:]
with timer('恢复模型和参数'):
train = trainer(is_recog=True)
if len(fname_or_list) == 0:
print('没有指定视频，使用所有测试集作为演示')
fname_or_list = train.batch_generator.test_fname_list
train.saver.restore(train.sess, train.model_fname)
recog(fname_or_list, train=train)
def main(restore=True):
with tools.timer('build model'):
train = trainer()
with tools.timer('train'):
train.train(restore=restore)
train.close()
