使用TensorFlow进行目标识别(二)——训练和识别

2018-02-12

经过上一节数据准备之后，开始进行配置训练和目标识别

1.配置object detection模型
下载Models源码
根据object detection安装指引进行配置安装

2.下载并配置COCO数据集预训练模型
下载官方文档中提到的COCO-pretrained Faster R-CNN with Resnet-101 model模型，解压之后，将model.ckpt开头的三个文件拷贝到训练目录下。
拷贝models/research/object_detection/samples/configs/faster_rcnn_resnet101_coco.config文件到训练目录下。
修改faster_rcnn_resnet101_coco.config文件：
1）将num_classes改为1，本例中只有一个分类
2）将所有PATH_TO_BE_CONFIGURED改为本机路径，共5处

3.训练
执行脚本进行训练

python object_detection/train.py \
--logtostderr \
--pipeline_config_path=/home/paiconor/object-detection/training/faster_rcnn_resnet101_coco.config \
--train_dir=/home/paiconor/object-detection/training/trainingresult

实践证明不能用MOHI做训练。。16G内存不够霍霍的。。

训练过程中可以使用TensorBoard进行监控，指令如下：

1	tensorboard --logdir=/home/paiconor/object-detection/training/trainingresult

通过TotalLoss查看训练情况，从整体上看TotalLoss曲线是收敛的，训练效果令人满意。

4.Freeze Model模型导出
查看模型实际的效果前，我们需要把训练的过程文件导出成.pb模型文件。
运行脚本

python object_detection/export_inference_graph.py \
--input_type=image_tensor \
--pipeline_config_path=/home/paiconor/object-detection/training/faster_rcnn_resnet101_coco.config \
--trained_checkpoint_prefix=/home/paiconor/object-detection/training/export/model.ckpt-2107 \
--output_directory=/home/paiconor/object-detection/training/output_inference_graph.pb

导出完成后，在output_directory下，会生成frozen_inference_graph.pb、model.ckpt.data-00000-of-00001、model.ckpt.meta、model.ckpt.data文件。

5.使用训练成果进行目标检测
编写目标检测脚本

import cv2
import numpy as np
import tensorflow as tf
from object_detection.utils import label_map_util
from object_detection.utils import visualization_utils as vis_util
class TOD(object):
    def __init__(self):
        self.PATH_TO_CKPT = r'/home/paiconor/object-detection/training/output_inference_graph.pb/frozen_inference_graph.pb'
        self.PATH_TO_LABELS = r'/home/paiconor/object-detection/training/hand.pbtxt'
        self.NUM_CLASSES = 1
        self.detection_graph = self._load_model()
        self.category_index = self._load_label_map()
    def _load_model(self):
        detection_graph = tf.Graph()
        with detection_graph.as_default():
            od_graph_def = tf.GraphDef()
            with tf.gfile.GFile(self.PATH_TO_CKPT, 'rb') as fid:
                serialized_graph = fid.read()
                od_graph_def.ParseFromString(serialized_graph)
                tf.import_graph_def(od_graph_def, name='')
        return detection_graph
    def _load_label_map(self):
        label_map = label_map_util.load_labelmap(self.PATH_TO_LABELS)
        categories = label_map_util.convert_label_map_to_categories(label_map,
                                                                    max_num_classes=self.NUM_CLASSES,
                                                                    use_display_name=True)
        category_index = label_map_util.create_category_index(categories)
        return category_index
    def detect(self, image):
        with self.detection_graph.as_default():
            with tf.Session(graph=self.detection_graph) as sess:
                # Expand dimensions since the model expects images to have shape: [1, None, None, 3] 
                image_np_expanded = np.expand_dims(image, axis=0)
                image_tensor = self.detection_graph.get_tensor_by_name('image_tensor:0')
                boxes = self.detection_graph.get_tensor_by_name('detection_boxes:0')
                scores = self.detection_graph.get_tensor_by_name('detection_scores:0')
                classes = self.detection_graph.get_tensor_by_name('detection_classes:0')
                num_detections = self.detection_graph.get_tensor_by_name('num_detections:0')
                # Actual detection. 
                (boxes, scores, classes, num_detections) = sess.run(
                    [boxes, scores, classes, num_detections],
                    feed_dict={image_tensor: image_np_expanded})
                # Visualization of the results of a detection. 
                vis_util.visualize_boxes_and_labels_on_image_array(
                    image,
                    np.squeeze(boxes),
                    np.squeeze(classes).astype(np.int32),
                    np.squeeze(scores),
                    self.category_index,
                    use_normalized_coordinates=True,
                    line_thickness=8)
        cv2.namedWindow("detection", cv2.WINDOW_NORMAL)
        cv2.imshow("detection", image)
        cv2.waitKey(0)
if __name__ == '__main__':
    image = cv2.imread('image.jpeg')
    detecotr = TOD()
    detecotr.detect(image)

识别效果：

目前对于纯色背景识别效果不错，但是如果背景比较复杂则会出现识别错误的情况，还需要继续优化。