`
rensanning
  • 浏览: 3547291 次
  • 性别: Icon_minigender_1
  • 来自: 大连
博客专栏
Efef1dba-f7dd-3931-8a61-8e1c76c3e39f
使用Titanium Mo...
浏览量:38123
Bbab2146-6e1d-3c50-acd6-c8bae29e307d
Cordova 3.x入门...
浏览量:607206
C08766e7-8a33-3f9b-9155-654af05c3484
常用Java开源Libra...
浏览量:682188
77063fb3-0ee7-3bfa-9c72-2a0234ebf83e
搭建 CentOS 6 服...
浏览量:89278
E40e5e76-1f3b-398e-b6a6-dc9cfbb38156
Spring Boot 入...
浏览量:401737
Abe39461-b089-344f-99fa-cdfbddea0e18
基于Spring Secu...
浏览量:69668
66a41a70-fdf0-3dc9-aa31-19b7e8b24672
MQTT入门
浏览量:91670
社区版块
存档分类
最新评论

TensorFlow 之 物体检测

 
阅读更多
谷歌开源的一套智能对象物体检测识别系统,让图像搜索和街景服务更准确。

相关文章:
TensorFlow 之 入门体验
TensorFlow 之 手写数字识别MNIST
TensorFlow 之 物体检测
TensorFlow 之 构建人物识别系统

(1)安装Protobuf
TensorFlow内部使用Protocol Buffers,物体检测需要特别安装一下。
# yum info protobuf protobuf-compiler
2.5.0 <-版本太低需要protobuf 2.6.1以上版本
# yum -y install autoconf automake libtool curl make g++ unzip
# cd /usr/local/src/
# wget https://github.com/google/protobuf/archive/v3.3.1.tar.gz -O protobuf-3.3.1.tar.gz
# tar -zxvf protobuf-3.3.1.tar.gz
# cd protobuf-3.3.1
# ./autogen.sh
# ./configure --prefix=/usr/local/protobuf
# make
# make install
# ldconfig
# export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/protobuf/lib
# export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/protobuf/lib
# export PATH=$PATH:/usr/local/protobuf/bin
# protoc --version
libprotoc 3.3.1


(2)配置Tensorflow物体检测API
# source /usr/local/tensorflow2/bin/activate
# cd /usr/local/tensorflow2/tensorflow-models


安装依赖包
# pip install pillow
# pip install lxml
# pip install jupyter
# pip install matplotlib


Protobuf编译
# protoc object_detection/protos/*.proto --python_out=.


设置环境变量
# export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/slim
# ldconfig


测试
# python object_detection/builders/model_builder_test.py

输出OK表示设置完成

(3)查看文档运行Demo
使用预训练模型来检测图像中的物体。官方提供了基于jupyter的教程。
# source /usr/local/tensorflow2/bin/activate
# cd /usr/local/tensorflow2/tensorflow-models/object_detection/
# jupyter notebook --generate-config --allow-root
# python -c 'from notebook.auth import passwd;print(passwd())'
Enter password:123456
Verify password:123456
sha1:7d026454901a:009ae34a09296674d4a13521b80b8527999fd3da
# vi /root/.jupyter/jupyter_notebook_config.py
c.NotebookApp.password = 'sha1:7d026454901a:009ae34a09296674d4a13521b80b8527999fd3da'
# jupyter notebook --ip=127.0.0.1 --allow-root


访问:http://127.0.0.1:8888/ 打开object_detection_tutorial.ipynb。
http://127.0.0.1:8888/notebooks/object_detection_tutorial.ipynb


默认是处理 object_detection/test_images 文件夹下的image1.jpg、image2.jpg,如果想识别其他图像可以把倒数第二个Cell的代码修改一下:
# TEST_IMAGE_PATHS = [ os.path.join(PATH_TO_TEST_IMAGES_DIR, 'image{}.jpg'.format(i)) for i in range(1, 3) ]
TEST_IMAGE_PATHS = ['<your image path>']


在最后一个cell里添加2行代码:
plt.figure(figsize=IMAGE_SIZE)
plt.imshow(image_np)

->
print(image_path.split('.')[0]+'_labeled.jpg') # Add
plt.figure(figsize=IMAGE_SIZE, dpi=300) # Modify
plt.imshow(image_np)
plt.savefig(image_path.split('.')[0] + '_labeled.jpg') # Add



然后从头到尾挨个执行每个Cell后等结果。(Download Model那部分代码需要从网上下载文件比较慢!)


执行完成后在object_detection/test_images 文件夹下就能看到结果图了。
image1_labeled.jpg
image2_labeled.jpg



比较一下官方提供的检测结果图,可见和机器于很大关系:



(4)编码检测图像

从ImageNet中取一张图2008_004037.jpg测试,然后把 object_detection_tutorial.ipynb 里的代码改成可直接运行代码

# vi object_detect_demo.py
# python object_detect_demo.py


import numpy as np
import os
import six.moves.urllib as urllib
import sys
import tarfile
import tensorflow as tf
import zipfile
import matplotlib

# Matplotlib chooses Xwindows backend by default.
matplotlib.use('Agg')

from collections import defaultdict
from io import StringIO
from matplotlib import pyplot as plt
from PIL import Image
from utils import label_map_util
from utils import visualization_utils as vis_util

##################### Download Model
# What model to download.
MODEL_NAME = 'ssd_mobilenet_v1_coco_11_06_2017'
MODEL_FILE = MODEL_NAME + '.tar.gz'
DOWNLOAD_BASE = 'http://download.tensorflow.org/models/object_detection/'

# Path to frozen detection graph. This is the actual model that is used for the object detection.
PATH_TO_CKPT = MODEL_NAME + '/frozen_inference_graph.pb'

# List of the strings that is used to add correct label for each box.
PATH_TO_LABELS = os.path.join('data', 'mscoco_label_map.pbtxt')

NUM_CLASSES = 90

# Download model if not already downloaded
if not os.path.exists(PATH_TO_CKPT):
    print('Downloading model... (This may take over 5 minutes)')
    opener = urllib.request.URLopener()
    opener.retrieve(DOWNLOAD_BASE + MODEL_FILE, MODEL_FILE)
    print('Extracting...')
    tar_file = tarfile.open(MODEL_FILE)
    for file in tar_file.getmembers():
        file_name = os.path.basename(file.name)
        if 'frozen_inference_graph.pb' in file_name:
            tar_file.extract(file, os.getcwd())
else:
    print('Model already downloaded.')

##################### Load a (frozen) Tensorflow model into memory.
print('Loading model...')
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='')

##################### Loading label map
print('Loading label map...')
label_map = label_map_util.load_labelmap(PATH_TO_LABELS)
categories = label_map_util.convert_label_map_to_categories(label_map, max_num_classes=NUM_CLASSES, use_display_name=True)
category_index = label_map_util.create_category_index(categories)

##################### Helper code
def load_image_into_numpy_array(image):
  (im_width, im_height) = image.size
  return np.array(image.getdata()).reshape(
      (im_height, im_width, 3)).astype(np.uint8)

##################### Detection
# Path to test image
TEST_IMAGE_PATH = 'test_images/2008_004037.jpg'

# Size, in inches, of the output images.
IMAGE_SIZE = (12, 8)

print('Detecting...')
with detection_graph.as_default():
  with tf.Session(graph=detection_graph) as sess:
    print(TEST_IMAGE_PATH)
    image = Image.open(TEST_IMAGE_PATH)
    image_np = load_image_into_numpy_array(image)
    image_np_expanded = np.expand_dims(image_np, axis=0)
    image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
    boxes = detection_graph.get_tensor_by_name('detection_boxes:0')
    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})
    # Print the results of a detection.
    print(scores)
    print(classes)
    print(category_index)
    # 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)
    print(TEST_IMAGE_PATH.split('.')[0]+'_labeled.jpg')
    plt.figure(figsize=IMAGE_SIZE, dpi=300)
    plt.imshow(image_np)
    plt.savefig(TEST_IMAGE_PATH.split('.')[0] + '_labeled.jpg')


检测结果(scores、classes、category_index)如下:
引用
[[ 0.91731095  0.80875194  0.67557526  0.67192227  0.3568708   0.23992854
   0.21897335  0.21443138  0.17383011  0.15901341  0.15674619  0.1558814
   0.15265906  0.1489363   0.14805503  0.13470834  0.132047    0.12655555
   0.12086334  0.11752894  0.10897312  0.10791111  0.10386674  0.10181901
   0.09687284  0.09644313  0.0929096   0.09187065  0.08420605  0.08250966
   0.08131051  0.07928694  0.07632151  0.07570603  0.0749495   0.07267584
   0.07258119  0.07075463  0.06964011  0.06901822  0.06894562  0.06892171
   0.06805679  0.06769397  0.06536105  0.06501643  0.06417865  0.06416738
   0.06377003  0.0634084   0.06247949  0.06245064  0.06173467  0.06126672
   0.06037482  0.05930964  0.05813492  0.05751488  0.05747007  0.05746768
   0.05737954  0.05694786  0.05581251  0.05559204  0.05539726  0.054422
   0.05410738  0.05389332  0.05359224  0.05349119  0.05328105  0.05284562
   0.0527565   0.05231072  0.05224103  0.05190464  0.05123441  0.05110639
   0.05002856  0.04982324  0.04956287  0.04943769  0.04906119  0.04891028
   0.04835404  0.04812568  0.0470486   0.04596276  0.04592303  0.04565331
   0.04564101  0.04550403  0.04531116  0.04507401  0.04495776  0.04489629
   0.04475424  0.0447024   0.04434219  0.04395287]]
[[  1.   1.  44.  44.  44.  44.  44.  75.  44.  44.  44.  82.  44.  88.
   79.  44.  44.  44.  88.  44.  88.  79.  44.  82.   1.  47.  88.  67.
   44.  70.  47.  79.  67.  67.  67.  67.  79.  72.  47.   1.  44.  44.
   44.   1.  67.  75.  72.  62.   1.   1.  44.  82.  79.  47.  79.  67.
   44.   1.  51.  75.  79.  51.  79.  62.  67.  44.  82.  82.  79.  82.
   79.  75.  72.  82.   1.   1.  46.  88.  82.  82.  82.  44.  67.  62.
   82.  79.  62.   1.  67.   1.  82.   1.  67.   1.  44.  88.  79.  51.
   44.  82.]]
{1: {'id': 1, 'name': u'person'}, 2: {'id': 2, 'name': u'bicycle'}, 3: {'id': 3, 'name': u'car'}, 4: {'id': 4, 'name': u'motorcycle'}, 5: {'id': 5, 'name': u'airplane'}, 6: {'id': 6, 'name': u'bus'}, 7: {'id': 7, 'name': u'train'}, 8: {'id': 8, 'name': u'truck'}, 9: {'id': 9, 'name': u'boat'}, 10: {'id': 10, 'name': u'traffic light'}, 11: {'id': 11, 'name': u'fire hydrant'}, 13: {'id': 13, 'name': u'stop sign'}, 14: {'id': 14, 'name': u'parking meter'}, 15: {'id': 15, 'name': u'bench'}, 16: {'id': 16, 'name': u'bird'}, 17: {'id': 17, 'name': u'cat'}, 18: {'id': 18, 'name': u'dog'}, 19: {'id': 19, 'name': u'horse'}, 20: {'id': 20, 'name': u'sheep'}, 21: {'id': 21, 'name': u'cow'}, 22: {'id': 22, 'name': u'elephant'}, 23: {'id': 23, 'name': u'bear'}, 24: {'id': 24, 'name': u'zebra'}, 25: {'id': 25, 'name': u'giraffe'}, 27: {'id': 27, 'name': u'backpack'}, 28: {'id': 28, 'name': u'umbrella'}, 31: {'id': 31, 'name': u'handbag'}, 32: {'id': 32, 'name': u'tie'}, 33: {'id': 33, 'name': u'suitcase'}, 34: {'id': 34, 'name': u'frisbee'}, 35: {'id': 35, 'name': u'skis'}, 36: {'id': 36, 'name': u'snowboard'}, 37: {'id': 37, 'name': u'sports ball'}, 38: {'id': 38, 'name': u'kite'}, 39: {'id': 39, 'name': u'baseball bat'}, 40: {'id': 40, 'name': u'baseball glove'}, 41: {'id': 41, 'name': u'skateboard'}, 42: {'id': 42, 'name': u'surfboard'}, 43: {'id': 43, 'name': u'tennis racket'}, 44: {'id': 44, 'name': u'bottle'}, 46: {'id': 46, 'name': u'wine glass'}, 47: {'id': 47, 'name': u'cup'}, 48: {'id': 48, 'name': u'fork'}, 49: {'id': 49, 'name': u'knife'}, 50: {'id': 50, 'name': u'spoon'}, 51: {'id': 51, 'name': u'bowl'}, 52: {'id': 52, 'name': u'banana'}, 53: {'id': 53, 'name': u'apple'}, 54: {'id': 54, 'name': u'sandwich'}, 55: {'id': 55, 'name': u'orange'}, 56: {'id': 56, 'name': u'broccoli'}, 57: {'id': 57, 'name': u'carrot'}, 58: {'id': 58, 'name': u'hot dog'}, 59: {'id': 59, 'name': u'pizza'}, 60: {'id': 60, 'name': u'donut'}, 61: {'id': 61, 'name': u'cake'}, 62: {'id': 62, 'name': u'chair'}, 63: {'id': 63, 'name': u'couch'}, 64: {'id': 64, 'name': u'potted plant'}, 65: {'id': 65, 'name': u'bed'}, 67: {'id': 67, 'name': u'dining table'}, 70: {'id': 70, 'name': u'toilet'}, 72: {'id': 72, 'name': u'tv'}, 73: {'id': 73, 'name': u'laptop'}, 74: {'id': 74, 'name': u'mouse'}, 75: {'id': 75, 'name': u'remote'}, 76: {'id': 76, 'name': u'keyboard'}, 77: {'id': 77, 'name': u'cell phone'}, 78: {'id': 78, 'name': u'microwave'}, 79: {'id': 79, 'name': u'oven'}, 80: {'id': 80, 'name': u'toaster'}, 81: {'id': 81, 'name': u'sink'}, 82: {'id': 82, 'name': u'refrigerator'}, 84: {'id': 84, 'name': u'book'}, 85: {'id': 85, 'name': u'clock'}, 86: {'id': 86, 'name': u'vase'}, 87: {'id': 87, 'name': u'scissors'}, 88: {'id': 88, 'name': u'teddy bear'}, 89: {'id': 89, 'name': u'hair drier'}, 90: {'id': 90, 'name': u'toothbrush'}}


获取前四个高于50%的物体结果如下:
引用
scores - 0.91731095  0.80875194  0.67557526  0.67192227
classes - 1.   1.  44.  44.
category_index - 1: {'id': 1, 'name': u'person'} 44: {'id': 44, 'name': u'bottle'}

图里也标出了【91%person、80%person、67%bottle、67%bottle】这四个物体:


(4)本地运行

1)生成 TFRecord
将jpg图片数据转换成TFRecord数据。
# cd /usr/local/tensorflow2/tensorflow-models/object_detection
# wget http://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz
# wget http://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz
# tar -zxvf annotations.tar.gz
# tar -zxvf images.tar.gz
# python create_pet_tf_record.py --data_dir=`pwd` --output_dir=`pwd`

images里全是已经标记好的jpg图片。执行完成后,会在当前目录下生成2个文件:pet_train.record、pet_val.record。

2)配置pipeline
在object_detection/samples下有各种模型的通道配置,复制一份出来用。
# wget http://download.tensorflow.org/models/object_detection/faster_rcnn_resnet101_coco_11_06_2017.tar.gz
# tar -zxvf faster_rcnn_resnet101_coco_11_06_2017.tar.gz
# cp samples/configs/faster_rcnn_resnet101_pets.config mypet.config
# vi mypet.config

修改PATH_TO_BE_CONFIGURED部分如下:
引用
fine_tune_checkpoint: "/usr/local/tensorflow2/tensorflow-models/object_detection/faster_rcnn_resnet101_coco_11_06_2017/model.ckpt"
from_detection_checkpoint: true

train_input_reader: {
  tf_record_input_reader {
    input_path: "/usr/local/tensorflow2/tensorflow-models/object_detection/pet_train.record"
  }
  label_map_path: "/usr/local/tensorflow2/tensorflow-models/object_detection/data/pet_label_map.pbtxt"
}

eval_input_reader: {
  tf_record_input_reader {
    input_path: "/usr/local/tensorflow2/tensorflow-models/object_detection/pet_val.record"
  }
  label_map_path: "/usr/local/tensorflow2/tensorflow-models/object_detection/data/pet_label_map.pbtxt"
}

from_detection_checkpoint设置为true,fine_tune_checkpoint需要设置检查点的路径。采用别人训练出来的checkpoint可以减少训练时间。
检查点的下载地址参考:
https://github.com/tensorflow/models/blob/master/object_detection/g3doc/detection_model_zoo.md

3)训练评估
# mkdir -p /usr/local/tensorflow2/tensorflow-models/object_detection/model/train
# mkdir -p /usr/local/tensorflow2/tensorflow-models/object_detection/model/eval


-- 训练 --
# python object_detection/train.py \
     --logtostderr \
     --pipeline_config_path='/usr/local/tensorflow2/tensorflow-models/object_detection/mypet.config' \
     --train_dir='/usr/local/tensorflow2/tensorflow-models/object_detection/model/train'

引用
INFO:tensorflow:Starting Session.
INFO:tensorflow:Saving checkpoint to path /usr/local/tensorflow2/tensorflow-models/object_detection/model/train/model.ckpt
INFO:tensorflow:Starting Queues.
INFO:tensorflow:global_step/sec: 0
INFO:tensorflow:Recording summary at step 0.


-- 评估 --
# python object_detection/eval.py \
    --logtostderr \
    --pipeline_config_path='/usr/local/tensorflow2/tensorflow-models/object_detection/mypet.config' \
    --checkpoint_dir='/usr/local/tensorflow2/tensorflow-models/object_detection/model/train' \
    --eval_dir='/usr/local/tensorflow2/tensorflow-models/object_detection/model/eval'

eval文件夹下会生成以下文件,一个文件对应一个image:
events.out.tfevents.1499152949.localhost.localdomain
events.out.tfevents.1499152964.localhost.localdomain
events.out.tfevents.1499152980.localhost.localdomain

-- 查看结果 --
# tensorboard --logdir=/usr/local/tensorflow/tensorflow-models/object_detection/model/


*** train和eval执行后直到终止命令前一直运行
*** 训练、评估、查看可以开3个终端分别同时运行

6月20号之前下载的tensorflow-models-master.zip是兼容Python3的会有很多问题:
https://github.com/tensorflow/models/issues/1597
https://github.com/tensorflow/models/pull/1614/files
比如:

引用
Traceback (most recent call last):
  File "create_pet_tf_record.py", line 213, in <module>
    tf.app.run()
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "create_pet_tf_record.py", line 208, in main
    image_dir, train_examples)
  File "create_pet_tf_record.py", line 177, in create_tf_record
    tf_example = dict_to_tf_example(data, label_map_dict, image_dir)
  File "create_pet_tf_record.py", line 131, in dict_to_tf_example
    'image/filename': dataset_util.bytes_feature(data['filename']),
  File "/usr/local/tensorflow/tensorflow-models/object_detection/utils/dataset_util.py", line 30, in bytes_feature
    return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
TypeError: 'leonberger_185.jpg' has type str, but expected one of: bytes


引用
Traceback (most recent call last):
  File "object_detection/train.py", line 198, in <module>
    tf.app.run()
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "object_detection/train.py", line 194, in main
    worker_job_name, is_chief, FLAGS.train_dir)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/trainer.py", line 184, in train
    data_augmentation_options)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/trainer.py", line 77, in _create_input_queue
    prefetch_queue_capacity=prefetch_queue_capacity)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/core/batcher.py", line 81, in __init__
    {key: tensor.get_shape() for key, tensor in tensor_dict.iteritems()})
AttributeError: 'dict' object has no attribute 'iteritems'


引用
Traceback (most recent call last):
  File "object_detection/train.py", line 198, in <module>
    tf.app.run()
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "object_detection/train.py", line 194, in main
    worker_job_name, is_chief, FLAGS.train_dir)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/trainer.py", line 184, in train
    data_augmentation_options)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/trainer.py", line 77, in _create_input_queue
    prefetch_queue_capacity=prefetch_queue_capacity)
  File "/usr/local/tensorflow/tensorflow-models/object_detection/core/batcher.py", line 93, in __init__
    num_threads=num_batch_queue_threads)
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/training/input.py", line 919, in batch
    name=name)
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/training/input.py", line 697, in _batch
    tensor_list = _as_tensor_list(tensors)
  File "/usr/local/tensorflow/lib/python3.6/site-packages/tensorflow/python/training/input.py", line 385, in _as_tensor_list
    return [tensors[k] for k in sorted(tensors)]
TypeError: '<' not supported between instances of 'tuple' and 'str'

等等
  • 大小: 28.9 KB
  • 大小: 24.3 KB
  • 大小: 1.2 MB
  • 大小: 1.5 MB
  • 大小: 1.3 MB
  • 大小: 30.3 KB
  • 大小: 364.2 KB
  • 大小: 377 KB
3
0
分享到:
评论
1 楼 lzfying 2017-11-17  
这些坑都遇到了

相关推荐

    实用代码 11 TensorFlow物体检测

    实用代码 11 TensorFlow物体检测实用代码 11 TensorFlow物体检测实用代码 11 TensorFlow物体检测实用代码 11 TensorFlow物体检测实用代码 11 TensorFlow物体检测实用代码 11 TensorFlow物体检测实用代码 11 ...

    tensorflow物体检测-图片目标分类计数

    利用tensorflow官方源码识别图片的基础上 删减不必要代码,加快识别 此外加了少量代码 统计图片中各类目标出现次数 有兴趣的可以在此基础上设计视频不同目标计数

    Tensorflow-物体检测-Faster-Rcnn解读 教学视频

    物体检测-Faster-Rcnn、原理+实战,你见过最接地气的课程。课程首先讲解物体检测的初期算法,对比不同效果与设计思想从而引入faster-rcnn三代算法,对三代算法原理进行详细解读。在学习阶段我们选择了tensorflow版本...

    tensorflow检测物体实例

    tensorflow检测物体实例:TensorFlow.js是Tensorflow的JavaScript版本。你可能听说过Tensorflow,因为它是世界上最受欢迎的机器学习工具。 TensorFlow.js充分吸取了机器学习的优点,并将应用于支持JavaScript 的node....

    tensorflow物体检测移植到Android端,实现物体检测自动拍照

    首先,我们需要准备的是一个训练好的TensorFlow物体检测模型。这个模型通常是基于SSD(Single Shot MultiBox Detector)或YOLO(You Only Look Once)等算法的预训练网络。这些模型能够实时地在图像中检测多个类别的...

    使用TensorFlow进行物体识别1

    在深度学习领域,TensorFlow作为一款开源机器学习框架,广泛应用于计算机视觉任务之中,尤其是物体识别方面。本文旨在通过一系列详细的操作步骤来指导读者如何利用TensorFlow进行物体识别。 #### 二、环境搭建与...

    Tensorflow-物体检测-Faster-Rcnn解读

    物体检测一些列模型视频讲解,tensorflow下的实现版本。

    tensorflow视频目标检测

    tensorflow视频目标检测 在官方tensorflow object detection api的基础上 削减繁杂多余的代码 实现摄像头实时读取 与识别物体 。下载本代码 需要选择模型pb文件与pbtxt文件 建议选择ssd模型 此外因电脑配置原因,...

    基于 TensorFlow 物体检测 API 的柔性对象识别系统.zip

    在本项目中,我们关注的是基于TensorFlow物体检测API实现的柔性对象识别系统。TensorFlow是一个强大的开源机器学习框架,广泛应用于深度学习领域,包括计算机视觉、自然语言处理等多种任务。柔性对象识别是计算机...

    Python实战TensorFlow物体检测(毕设 + 课设).zip

    项目中的1.txt文件可能是介绍项目背景和具体步骤的文档,而11 TensorFlow物体检测可能是更具体的教程或代码示例。通过这些资源,你应该能够逐步掌握物体检测技术,并完成相关的毕业设计或课程设计项目。记得在实践...

    谷歌Tensorflow 物体识别模块

    在给定的“物体识别模块”中,我们主要关注的是TensorFlow在计算机视觉中的应用,特别是物体检测这一子领域。 物体检测是计算机视觉的一个关键任务,它不仅要求识别图像中的对象,还要精确地定位出对象的位置。...

    TensorFlow物体检测

    **TensorFlow物体检测** 在深度学习领域,物体检测是一项关键任务,它涉及到识别图像或视频中的特定对象并确定它们的位置。TensorFlow,一个强大的开源机器学习框架,提供了多种实现物体检测的方法。本教程将深入...

    基于TensorFlow Lite物体识别安卓APP的设计与实现.zip

    在本项目中,我们主要探讨的是如何利用TensorFlow Lite这一轻量级的机器学习框架,在Android平台上设计并实现一个物体识别应用。TensorFlow Lite是Google为移动和嵌入式设备优化的TensorFlow版本,它允许开发者将...

    基于TensorFlow实现CNN水果检测.zip

    在本项目"基于TensorFlow实现CNN水果检测"中,我们主要探讨了如何利用深度学习框架TensorFlow构建卷积神经网络(CNN)模型来识别不同类型的水果。深度学习,特别是CNN,已经成为计算机视觉领域的重要工具,它能有效...

    基于tensorflow+anaconda的物体检测+基于Pycharm

    6. **11 TensorFlow物体检测**:这个文件名可能指代了一个特定的模块或者代码文件,其中包含了物体检测的实现细节。可能包括了加载预训练模型、图像预处理、模型推理以及结果可视化等步骤。 在实际操作中,首先需要...

    该项目基于Tensorflow API实现物体实时检测功能.zip

    其灵活性和易用性使得TensorFlow成为学术界和工业界的首选平台,用于实现包括物体检测在内的各种机器学习任务。 二、物体检测技术 物体检测是计算机视觉的重要分支,它的目标是在图像中识别并定位出多个不同类别的...

    tensorflow物体识别object detection官方预训练模型

    本篇文章将详细探讨“tensorflow物体识别object detection官方预训练模型”,以及如何利用这些模型进行物体识别。 首先,TensorFlow Object Detection API 是一个开源框架,它提供了丰富的工具和模型,用于构建、...

    30秒轻松实现TensorFlow物体检测

    TensorFlow物体检测是利用深度学习框架TensorFlow进行图像中物体识别的技术。本文将介绍如何利用TensorFlow提供的API和预训练模型来实现快速的物体检测功能。 首先,Google官方发布的TensorFlow物体检测API为我们...

Global site tag (gtag.js) - Google Analytics