如何在Jetson Nano上运行TensorFlow对象检测模型

基于http://bbs.gpuworld.cn/index.php?topic=73098.msg84029来做

问题1：download_detection_model及build_detection_graph API找不到定义
不清楚tensorflow对象检测API里面是否有这两个函数，使用find找不到，解决方法，使用tf_trt_models代替

问题2：在jetson nano上导入tf-trt优化后的graph时，出现以下错误“tensorflow.python.framework.errors_impl.NotFoundError: Op type not registered ‘TRTEngineOp’ in binary running on jetson-nano.”
原因是trt未import，加入"import tensorflow.contrib.tensorrt as trt"即可

问题3：基于mobilenet+SSD v1的推理速度在100ms~170ms之间
将jetson nano的用户图形界面关闭后，一帧画面的推理速度可以到50~60ms

/*关闭用户图形界面*/
sudo systemctl set-default multi-user.target
sudo reboot

/*开启用户图形界面*/
sudo systemctl set-default graphical.target
sudo reboot

/******************create_trt_graph.py***************************/
from tf_trt_models.detection import download_detection_model
from tf_trt_models.detection import build_detection_graph
import tensorflow.contrib.tensorrt as trt

config_path, checkpoint_path = download_detection_model('ssd_mobilenet_v1_coco')
frozen_graph, input_names, output_names = build_detection_graph(
            config=config_path,
            checkpoint=checkpoint_path,
            score_threshold=0.3,
            batch_size=1
            )
trt_graph = trt.create_inference_graph(
            input_graph_def=frozen_graph,
            outputs=output_names,
            max_batch_size=1,
            max_workspace_size_bytes=1 << 25,
            precision_mode='FP16',
            minimum_segment_size=50
            )

with open('./data/trt_graph.pb', 'wb') as f:
    f.write(trt_graph.SerializeToString())
    f.close()

import tensorflow as tf
import cv2
from IPython.display import Image as DisplayImage
import tensorflow.contrib.tensorrt as trt
import numpy as np
import os
import time
from PIL import Image

def get_frozen_graph(graph_file):
    """Read Frozen Graph file from disk."""
    with tf.gfile.FastGFile(graph_file, "rb") as f:
         graph_def = tf.GraphDef()
         graph_def.ParseFromString(f.read())
    return graph_def

# The TensorRT inference graph file downloaded from Colab or your local machine.
pb_fname = "./data/trt_graph.pb"
trt_graph = get_frozen_graph(pb_fname)

input_names = ['image_tensor']

# Create session and load graph
tf_config = tf.ConfigProto()
tf_config.gpu_options.allow_growth = True
#tf_sess = tf.Session(config=tf_config)
with tf.Session(config=tf_config) as tf_sess:
    tf.contrib.resampler
    tf.import_graph_def(trt_graph, name='')

    tf_input = tf_sess.graph.get_tensor_by_name(input_names[0] + ':0')
    tf_scores = tf_sess.graph.get_tensor_by_name('detection_scores:0')
    tf_boxes = tf_sess.graph.get_tensor_by_name('detection_boxes:0')
    tf_classes = tf_sess.graph.get_tensor_by_name('detection_classes:0')
    tf_num_detections = tf_sess.graph.get_tensor_by_name('num_detections:0')

    #IMAGE_PATH = "./data/image1.jpg"
    #image = cv2.imread(IMAGE_PATH)
    video_path = "./data/road.mp4"
    vid = cv2.VideoCapture(video_path)
    while True:
        return_value, frame = vid.read()
        if return_value:
            frame = cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
        else:
            raise ValueError("No Image!")
        image = cv2.resize(frame, (300, 300))

        prev_time = time.time()

        scores, boxes, classes, num_detections = tf_sess.run([tf_scores, tf_boxes, tf_classes, tf_num_detections], feed_dict={tf_inp
ut: image[None, ...]})
        boxes = boxes[0]  # index by 0 to remove batch dimension
        scores = scores[0]
        classes = classes[0]
        num_detections = int(num_detections[0])

        curr_time = time.time()
        exec_time = curr_time - prev_time
        print("time:%.2f ms" % (1000 * exec_time))

        # Boxes unit in pixels (image coordinates).
        boxes_pixels = []
        for i in range(num_detections):
            # scale box to image coordinates
            box = boxes[i] * np.array([image.shape[0],image.shape[1], image.shape[0], image.shape[1]])
            box = np.round(box).astype(int)
            boxes_pixels = np.append(boxes_pixels,box)

        boxes_pixels = boxes_pixels.reshape(int(len(boxes_pixels)/4),4)
#        print("boxes_pixels")
#        print(boxes_pixels)
#        print("scores:")
#        print(scores[0])
#        print("classes")
#        print(classes[0])
        # Remove overlapping boxes with non-max suppression, return picked indexes.
        pick = tf_sess.run(tf.image.non_max_suppression(boxes=boxes_pixels,scores=scores[:num_detections],iou_threshold=0.1,max_outp
ut_size=5))
        print(pick)

        for i in pick:
            box = boxes_pixels[i]
            box = np.round(box).astype(int)
#            print("box[%d] (%d,%d,%d,%d)" % (i,box[1],box[0],box[3],box[2]))
            # Draw bounding box.
  image = cv2.rectangle(image, (box[1], box[0]), (box[3], box[2]), (0, 255, 0), 2)
            label = "{}:{:.2f}".format(int(classes[i]), scores[i])
            # Draw label (class index and probability).
            #draw_label(image, (box[1], box[0]), label)
            cv2.imwrite("./data/result.jpg",image)