小试tensorflow的目标识别

先看效果吧
- 乖乖女儿
- 笨拙滑冰
代码
结束

利用tensorflow提供的目标识别的库，做图像识别简直是轻松简单酷。（google大法好）

先看效果吧

乖乖女儿

笨拙滑冰

代码

训练模型

首先需要安装“插件”：tensorflow/model/object_detection

tensorflow官方提供了如下几种训练模型，我们选择速度最快的“ssd_mobilenet_v1_coco_11_06_2017”：

首先是加载计算图graph。一个TensorFlow的运算，被表示为一个数据流的图。一幅图中包含一些操作（Operation）对象，这些对象是计算节点。

global detection_graph
detection_graph = tf.Graph()
with detection_graph.as_default():
    od_graph_def = tf.GraphDef()
    with tf.gfile.GFile(PATH_TO_CKPT, 'rb') as fid:
        serialized_graph = fid.read()
        od_graph_def.ParseFromString(serialized_graph)
        tf.import_graph_def(od_graph_def, name='')

然后是加载识别标签，标签如下图：

识别图片

需要先用numpy库把图片转位“矩阵”，把图片矩阵和标签投入session中运算，“很简单”地得到识别到的box和类别及概率，将这些信息附加到原始图片上：

def detect_object(image_np, sess, detection_graph, category_index):
    # Expand dimensions since the model expects images to have shape: [1, None, None, 3]
    image_np_expanded = np.expand_dims(image_np, axis=0)
    image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
    # Each box represents a part of the image where a particular object was detected.
    boxes = detection_graph.get_tensor_by_name('detection_boxes:0')
    # Each score represent how level of confidence for each of the objects.
    # Score is shown on the result image, together with the class label.
    scores = detection_graph.get_tensor_by_name('detection_scores:0')
    classes = detection_graph.get_tensor_by_name('detection_classes:0')
    num_detections = 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,
        np.squeeze(boxes),
        np.squeeze(classes).astype(np.int32),
        np.squeeze(scores),
        category_index,
        use_normalized_coordinates=True,
        line_thickness=8)    
    return image_np

识别视频

以上只是对图片的识别，下一步需要用到moviepy库。对视频文件进行切片得到图片列表，分别对这些图片进行识别后再组装为视频：

white_output = PATH_TO_TEST_VIDEOS_DIR + '/skate_out.mp4'
clip = VideoFileClip(os.path.join(PATH_TO_TEST_VIDEOS_DIR, 'skate.mp4')).subclip(0,10)
white_clip = clip.fl_image(process_image)
white_clip.write_videofile(white_output, audio=False)

结束

完整代码在https://github.com/baixiaoustc/tensorflow_objectdetection

参考：

https://github.com/tensorflow/models/blob/master/object_detection/object_detection_tutorial.ipynb

https://github.com/priya-dwivedi/Deep-Learning/blob/master/Object_Detection_Tensorflow_API.ipynb

起飞就起飞

目录