def really_commands(mode):
modules = utils.get_classes('workflow.{0}'.format(mode))
for module in modules:
# get RCE if you can edit general file in workflow folder :)
module_name = module[0].strip()
module_object = eval('{0}.{1}'.format(mode, module_name))
# parsing commands
try:
routines = module_object.commands
except:
continue
for routine, commands in routines.items():
for command in commands:
item = command
item['mode'] = mode
item['speed'] = routine
item['module'] = module_name
item['alias'] = module_name + "__" + routine.lower() + "__" + \
str(item.get('banner')).lower()
Commands.objects.create(**item)
reports = module_object.reports
parse_report(reports, module_name, mode)
# File name : nms_demo.py
# Author : YunYang1994
# Created date: 2018-11-27 13:02:17
# Description :
#
#================================================================
import time
import numpy as np
import tensorflow as tf
from PIL import Image
from core import utils
SIZE = [416, 416]
# SIZE = [608, 608]
classes = utils.get_classes('./data/coco.names')
num_classes = len(classes)
img = Image.open('./data/demo_data/611.jpg')
img_resized = np.array(img.resize(size=tuple(SIZE)), dtype=np.float32)
img_resized = img_resized / 255.
cpu_nms_graph, gpu_nms_graph = tf.Graph(), tf.Graph()
# nms on GPU
input_tensor, output_tensors = utils.read_pb_return_tensors(
gpu_nms_graph, "./checkpoint/yolov3_gpu_nms.pb",
["Placeholder:0", "concat_10:0", "concat_11:0", "concat_12:0"])
with tf.Session(graph=gpu_nms_graph) as sess:
for i in range(5):
start = time.time()
boxes, scores, labels = sess.run(
output_tensors,
config.gpu_options.allocator_type = 'BFC' # A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
config.gpu_options.per_process_gpu_memory_fraction = 0.8
config.gpu_options.allow_growth = True
set_session(tf.Session(config=config))
if __name__ == "__main__":
# 训练后的模型保存路径
log_dir = os.path.join(cfg.PATH.logs, 'v4')
# 权值文件
weights_path = cfg.PATH.weight_path
# 输入的shape大小
input_shape = cfg.TRAIN.input_size
# 是否对损失进行归一化
normalize = True
# 获取classes和anchor
class_names = get_classes(cfg.PATH.classes_info)
anchors = get_anchors(cfg.PATH.anchors_info)
# 一共有多少类和多少先验框
num_classes = len(class_names)
num_anchors = len(anchors)
#------------------------------------------------------#
# Yolov4的tricks应用
# mosaic 马赛克数据增强 True or False
# 实际测试时mosaic数据增强并不稳定，所以默认为False
# Cosine_scheduler 余弦退火学习率 True or False
# label_smoothing 标签平滑 0.01以下一般 如0.01、0.005
#------------------------------------------------------#
mosaic = False
Cosine_scheduler = False
label_smoothing = 0.005