使用ssd mobilenet_v1

Question

我正在使用tensorflow-gpu和mobilenet ssd v1开发对象检测工具。它运行成功，但是使用opencv videoCapture的速度大约为（1 FPS）。 GPU使用率仅显示1-2％。请提出任何解决方案。

• 系统：
• Windows 10
• CUDA 9.0
• cudnn 7.3.1
• tensorflow-gpu 1.12.0

脚本初始化输出

counting_started2018-12-20 13:52:29.390097: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
2018-12-20 13:52:30.345891: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: GeForce GTX 1050 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.62
pciBusID: 0000:01:00.0
totalMemory: 4.00GiB freeMemory: 3.30GiB
2018-12-20 13:52:30.346365: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 13:52:31.688196: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 13:52:31.688390: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 13:52:31.688504: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 13:52:31.688754: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 3013 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1050 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1)
2018-12-20 13:52:35.488701: I tensorflow/stream_executor/dso_loader.cc:151] successfully opened CUDA library cupti64_90.dll locally

detection.py

with detection_graph.as_default():
    with tf.device('/gpu:0'):
        with tf.Session(graph=detection_graph) as sess:

            d_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.
            d_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.
            d_scores = detection_graph.get_tensor_by_name('detection_scores:0')
            d_classes = detection_graph.get_tensor_by_name('detection_classes:0')
            d_num_detections = detection_graph.get_tensor_by_name('num_detections:0')

            while True:

                import gc

                gc.collect()

                ret, image_np = cap.read()

                if image_np is None:
                    break

                image_rgb = cv2.cvtColor(image_np, cv2.COLOR_BGR2RGB)

                (height, width) = image_np.shape[:2]

                rects = []

                if totalFrames % 1 == 0:

                    trackers = []

                    status = "Detecting"

                    image_np_expanded = np.expand_dims(image_np, axis=0)

                    options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
                    run_metadata = tf.RunMetadata()

                    # Actual detection.
                    (boxes, scores, classes, num_detections) = sess.run(
                        [d_boxes, d_scores, d_classes, d_num_detections],
                        feed_dict={d_image_tensor: image_np_expanded},
                        options=options, run_metadata=run_metadat


                   # 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=2)

                    cv2.imshow('object detection', image_np)
                    if cv2.waitKey(25) == ord('q'):
                        cv2.destroyAllWindows()
                        break

在上图中看到时，它将显示GPU的实际使用情况。在运行时始终为1-2％。我想念什么？