Tensorflow Object Detection API使用TFRecord文件保存正负样本数据。下面是一个包含代码示例的解决方法:
首先,需要准备好正负样本的图像,并将它们分为训练集和验证集。
接下来,将正负样本图像转换为Tensorflow Object Detection API所需的TFRecord格式。可以使用以下代码示例来完成此操作:
import tensorflow as tf
from object_detection.utils import dataset_util
def create_tf_example(example):
height = example['height']
width = example['width']
filename = example['filename']
image_format = b'jpg'
encoded_image_data = tf.io.gfile.GFile(example['path'], 'rb').read()
xmins = []
xmaxs = []
ymins = []
ymaxs = []
classes_text = []
classes = []
for box in example['boxes']:
xmins.append(box['xmin'] / width)
xmaxs.append(box['xmax'] / width)
ymins.append(box['ymin'] / height)
ymaxs.append(box['ymax'] / height)
classes_text.append(box['label'].encode('utf8'))
classes.append(box['class_id'])
tf_example = tf.train.Example(features=tf.train.Features(feature={
'image/height': dataset_util.int64_feature(height),
'image/width': dataset_util.int64_feature(width),
'image/filename': dataset_util.bytes_feature(filename.encode('utf8')),
'image/source_id': dataset_util.bytes_feature(filename.encode('utf8')),
'image/encoded': dataset_util.bytes_feature(encoded_image_data),
'image/format': dataset_util.bytes_feature(image_format),
'image/object/bbox/xmin': dataset_util.float_list_feature(xmins),
'image/object/bbox/xmax': dataset_util.float_list_feature(xmaxs),
'image/object/bbox/ymin': dataset_util.float_list_feature(ymins),
'image/object/bbox/ymax': dataset_util.float_list_feature(ymaxs),
'image/object/class/text': dataset_util.bytes_list_feature(classes_text),
'image/object/class/label': dataset_util.int64_list_feature(classes),
}))
return tf_example
def create_tf_record(output_path, examples):
writer = tf.io.TFRecordWriter(output_path)
for example in examples:
tf_example = create_tf_example(example)
writer.write(tf_example.SerializeToString())
writer.close()
# 示例数据
examples = [
{
'path': 'path/to/image1.jpg',
'filename': 'image1.jpg',
'height': 480,
'width': 640,
'boxes': [
{
'xmin': 100,
'xmax': 200,
'ymin': 150,
'ymax': 250,
'label': 'person',
'class_id': 1
},
{
'xmin': 300,
'xmax': 400,
'ymin': 200,
'ymax': 300,
'label': 'dog',
'class_id': 2
}
]
},
# 添加更多示例数据...
]
# 创建TFRecord文件
create_tf_record('path/to/train.record', examples)
以上代码会将示例数据转换为TFRecord格式,并将其保存为train.record
文件。
val.record
文件。使用以上方法可以将正负样本数据保存为Tensorflow Object Detection API所需的TFRecord格式,以便在训练模型时使用。
下一篇:保存在XML中的隐藏字符