SSD目標檢測實戰（TF項目）——VOC2007

TF項目實戰（SSD目標檢測）-VOC2007

理論詳解：https://blog.csdn.net/u013989576/article/details/73439202

訓練好的模型和代碼會公布在網上（含 VOC數據集  vgg16 模型  以及訓練好的模型）：

 待續

步驟：

1.代碼地址：https://github.com/balancap/SSD-Tensorflow

2.解壓ssd_300_vgg.ckpt.zip 到checkpoint文件夾下（另外將vgg16模型放在本路徑下）

3.測試一下看看，在notebooks文件夾下創建demo_test.py,其實就是復制ssd_notebook.ipynb中的代碼，該py文件是完成對於單張圖片的測試。

import os
import math
import random

import numpy as np
import tensorflow as tf
import cv2

slim = tf.contrib.slim
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import sys

sys.path.append('../')
from nets import ssd_vgg_300, ssd_common, np_methods
from preprocessing import ssd_vgg_preprocessing
from notebooks import visualization

# TensorFlow session: grow memory when needed. TF, DO NOT USE ALL MY GPU MEMORY!!!
gpu_options = tf.GPUOptions(allow_growth=True)
config = tf.ConfigProto(log_device_placement=False, gpu_options=gpu_options)
isess = tf.InteractiveSession(config=config)
# Input placeholder.
net_shape = (300, 300)
data_format = 'NHWC'
img_input = tf.placeholder(tf.uint8, shape=(None, None, 3))
# Evaluation pre-processing: resize to SSD net shape.
image_pre, labels_pre, bboxes_pre, bbox_img = ssd_vgg_preprocessing.preprocess_for_eval(
    img_input, None, None, net_shape, data_format, resize=ssd_vgg_preprocessing.Resize.WARP_RESIZE)
image_4d = tf.expand_dims(image_pre, 0)

# Define the SSD model.
reuse = True if 'ssd_net' in locals() else None
ssd_net = ssd_vgg_300.SSDNet()
with slim.arg_scope(ssd_net.arg_scope(data_format=data_format)):
    predictions, localisations, _, _ = ssd_net.net(image_4d, is_training=False, reuse=reuse)

# Restore SSD model.
ckpt_filename = '../checkpoints/ssd_300_vgg.ckpt'
# ckpt_filename = '../checkpoints/VGG_VOC0712_SSD_300x300_ft_iter_120000.ckpt'
isess.run(tf.global_variables_initializer())
saver = tf.train.Saver()
saver.restore(isess, ckpt_filename)

# SSD default anchor boxes.
ssd_anchors = ssd_net.anchors(net_shape)


# Main image processing routine.
def process_image(img, select_threshold=0.5, nms_threshold=.45, net_shape=(300, 300)):
    # Run SSD network.
    rimg, rpredictions, rlocalisations, rbbox_img = isess.run([image_4d, predictions, localisations, bbox_img],
                                                              feed_dict={img_input: img})

    # Get classes and bboxes from the net outputs.
    rclasses, rscores, rbboxes = np_methods.ssd_bboxes_select(
        rpredictions, rlocalisations, ssd_anchors,
        select_threshold=select_threshold, img_shape=net_shape, num_classes=21, decode=True)

    rbboxes = np_methods.bboxes_clip(rbbox_img, rbboxes)
    rclasses, rscores, rbboxes = np_methods.bboxes_sort(rclasses, rscores, rbboxes, top_k=400)
    rclasses, rscores, rbboxes = np_methods.bboxes_nms(rclasses, rscores, rbboxes, nms_threshold=nms_threshold)
    # Resize bboxes to original image shape. Note: useless for Resize.WARP!
    rbboxes = np_methods.bboxes_resize(rbbox_img, rbboxes)
    return rclasses, rscores, rbboxes


# Test on some demo image and visualize output.
# 測試的文件夾
path = '../demo/'
image_names = sorted(os.listdir(path))
# 文件夾中的第幾張圖，-1代表最后一張
img = mpimg.imread(path + image_names[-1])
rclasses, rscores, rbboxes = process_image(img)

# visualization.bboxes_draw_on_img(img, rclasses, rscores, rbboxes, visualization.colors_plasma)
visualization.plt_bboxes(img, rclasses, rscores, rbboxes)

4. 將自己的數據集或者 VOC2007直接放在工程目錄下

5. 修改datasets文件夾中pascalvoc_common.py文件，將訓練類修改別成自己的（這里如果自己的類別）  本文以兩類為例子

VOC_LABELS = {
    'none': (0, 'Background'),
    'aeroplane': (1, 'Vehicle'),
    'bicycle': (2, 'Vehicle'),
    'bird': (3, 'Animal'),
    'boat': (4, 'Vehicle'),
    'bottle': (5, 'Indoor'),
    'bus': (6, 'Vehicle'),
    'car': (7, 'Vehicle'),
    'cat': (8, 'Animal'),
    'chair': (9, 'Indoor'),
    'cow': (10, 'Animal'),
    'diningtable': (11, 'Indoor'),
    'dog': (12, 'Animal'),
    'horse': (13, 'Animal'),
    'motorbike': (14, 'Vehicle'),
    'person': (15, 'Person'),
    'pottedplant': (16, 'Indoor'),
    'sheep': (17, 'Animal'),
    'sofa': (18, 'Indoor'),
    'train': (19, 'Vehicle'),
    'tvmonitor': (20, 'Indoor'),
}
#自己的數據
# VOC_LABELS = {
#     'none': (0, 'Background'),
#     'aeroplane': (1, 'Vehicle'),
# }

6.  將圖像數據轉換為tfrecods格式，修改datasets文件夾中的pascalvoc_to_tfrecords.py文件，然后更改文件的83行讀取方式為’rb‘，如果你的文件不是.jpg格式，也可以修改圖片的類型。

另外這個修改

7.運行tf_convert_data.py文件，但是需要傳給它一些參數：  這個文件生成TFrecords文件的代碼

但是該文件需要像類似於linux 命令那樣傳入參數。  pycharm中如何解決呢？？？

假設我們需要執行：python ./tf_convert_data.py   --dataset_name=pascalvoc  --dataset_dir=./VOC2007/  --output_name=voc_2007_train  --output_dir=./tfrecords_怎么辦呢？

 

我們可以在  run中的 Edit ... 進入

一、

二、

三、

參數：--dataset_name=pascalvoc --dataset_dir=./VOC2007/ --output_name=voc_2007_train --output_dir=./tfrecords_

然后執行該py文件就ok。

 

如果出現錯誤（文件夾相關的錯誤），則在工程下建立一個文件夾就可以了。

 

8.訓練模型train_ssd_network.py文件中修改

 

None代表一直訓練。

 

 

其它需要修改的文件：

① nets/ssd_vgg_300.py  (因為使用此網絡結構) ，修改87 和88行的類別

② train_ssd_network.py,修改類別120行，GPU占用量，學習率，batch_size等

③ eval_ssd_network.py 修改類別，66行

④ datasets/pascalvoc_2007.py 根據自己的訓練數據修改整個文件

 

9.開始訓練

 類似於第7步中的 三

訓練的主文件為  train_ssd_network.py

參數為：

--train_dir=./train_model/ --dataset_dir=./tfrecords_/ --dataset_name=pascalvoc_2007 --dataset_split_name=train --model_name=ssd_300_vgg --checkpoint_path=./checkpoints/vgg_16.ckpt --checkpoint_model_scope=vgg_16 --checkpoint_exclude_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box --trainable_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box --save_summaries_secs=60 --save_interval_secs=600 --weight_decay=0.0005 --optimizer=adam --learning_rate=0.001 --learning_rate_decay_factor=0.94 --batch_size=24 --gpu_memory_fraction=0.9

 

訓練過程

：

 

 

10 測試：

先看效果

 

另外我修改了demo_test文件 調取電腦攝像投來執行代碼。

 

如果單獨看一張圖的效果則執行函數：

代碼如下：

__author__ = "WSX"
import os
import math
import random

import numpy as np
import tensorflow as tf
import cv2

slim = tf.contrib.slim
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import sys

sys.path.append('../')
from nets import ssd_vgg_300, ssd_common, np_methods
from preprocessing import ssd_vgg_preprocessing
from notebooks import visualization

# TensorFlow session: grow memory when needed. TF, DO NOT USE ALL MY GPU MEMORY!!!
gpu_options = tf.GPUOptions(allow_growth=True)
config = tf.ConfigProto(log_device_placement=False, gpu_options=gpu_options)
isess = tf.InteractiveSession(config=config)
# Input placeholder.
net_shape = (300, 300)
data_format = 'NHWC'
img_input = tf.placeholder(tf.uint8, shape=(None, None, 3))
# Evaluation pre-processing: resize to SSD net shape.
image_pre, labels_pre, bboxes_pre, bbox_img = ssd_vgg_preprocessing.preprocess_for_eval(
    img_input, None, None, net_shape, data_format, resize=ssd_vgg_preprocessing.Resize.WARP_RESIZE)
image_4d = tf.expand_dims(image_pre, 0)

# Define the SSD model.
reuse = True if 'ssd_net' in locals() else None
ssd_net = ssd_vgg_300.SSDNet()
with slim.arg_scope(ssd_net.arg_scope(data_format=data_format)):
    predictions, localisations, _, _ = ssd_net.net(image_4d, is_training=False, reuse=reuse)

# Restore SSD model.
ckpt_filename = '../checkpoints/ssd_300_vgg.ckpt'
# ckpt_filename = '../checkpoints/VGG_VOC0712_SSD_300x300_ft_iter_120000.ckpt'
isess.run(tf.global_variables_initializer())
saver = tf.train.Saver()
saver.restore(isess, ckpt_filename)

# SSD default anchor boxes.
ssd_anchors = ssd_net.anchors(net_shape)


# Main image processing routine.
def process_image(img, select_threshold=0.5, nms_threshold=.45, net_shape=(300, 300)):
    # Run SSD network.
    rimg, rpredictions, rlocalisations, rbbox_img = isess.run([image_4d, predictions, localisations, bbox_img],
                                                              feed_dict={img_input: img})

    # Get classes and bboxes from the net outputs.
    rclasses, rscores, rbboxes = np_methods.ssd_bboxes_select(
        rpredictions, rlocalisations, ssd_anchors,
        select_threshold=select_threshold, img_shape=net_shape, num_classes=21, decode=True)

    rbboxes = np_methods.bboxes_clip(rbbox_img, rbboxes)
    rclasses, rscores, rbboxes = np_methods.bboxes_sort(rclasses, rscores, rbboxes, top_k=400)
    rclasses, rscores, rbboxes = np_methods.bboxes_nms(rclasses, rscores, rbboxes, nms_threshold=nms_threshold)
    # Resize bboxes to original image shape. Note: useless for Resize.WARP!
    rbboxes = np_methods.bboxes_resize(rbbox_img, rbboxes)
    return rclasses, rscores, rbboxes


#===========================================測試部分===========================================
#----------------------------------單張圖片測試---------------------------
# Test on some demo image and visualize output.
# 測試的文件夾
def demo():
    path = '../demo/'
    image_names = sorted(os.listdir(path))
    # 文件夾中的第幾張圖，-1代表最后一張
    img = mpimg.imread(path + image_names[-1])
    print(img.shape)
    rclasses, rscores, rbboxes = process_image(img)

    # visualization.bboxes_draw_on_img(img, rclasses, rscores, rbboxes, visualization.colors_plasma)
    visualization.plt_bboxes(img, rclasses, rscores, rbboxes)


#======================================做成實時顯示的代碼===================================================
L = ["None","aeroplane","bicycle","bird ","boat ","bottle ","bus ","car ","cat ","chair","cow ","diningtable","dog","horse","motorbike","person",
     "pottedplant","sheep","sofa","train","tvmonitor"]
colors_tableau = [(255, 255, 255), (31, 119, 180), (174, 199, 232), (255, 127, 14), (255, 187, 120),
                  (44, 160, 44), (152, 223, 138), (214, 39, 40), (255, 152, 150),
                  (148, 103, 189), (197, 176, 213), (140, 86, 75), (196, 156, 148),
                  (227, 119, 194), (247, 182, 210), (127, 127, 127), (199, 199, 199),
                  (188, 189, 34), (219, 219, 141), (23, 190, 207), (158, 218, 229)]

def Load_video_show():  #獲取視頻
    video = cv2.VideoCapture("1.mp4")  # 0 表示攝像頭 ，   如果為文件路徑則 為加載視頻
    while (True):
        ret, frame = video.read()  #frame為 幀  這里當做一張圖
        frame = cv2.flip( frame ,1)  #鏡像變換，圖像正與不正
        cv2.resizeWindow("video", 640, 360)                #設置窗口大小
        frame = cv2.resize(frame, (640, 360))               #設置圖大小
        #cv2.imshow("video" ,frame)
        rclasses, rscores, rbboxes = process_image(frame)
        height = frame.shape[0]
        width = frame.shape[1]
        colors = dict()
        for i in range(rclasses.shape[0]):
            cls_id = int(rclasses[i])
            if cls_id >= 0:
                score = rscores[i]
                if cls_id not in colors:
                    colors[cls_id] = (random.random(), random.random(), random.random())
                ymin = int(rbboxes[i, 0] * height)
                xmin = int(rbboxes[i, 1] * width)
                ymax = int(rbboxes[i, 2] * height)
                xmax = int(rbboxes[i, 3] * width)
                cv2.rectangle(frame, (xmin, ymin), (xmax,ymax), colors[cls_id], 2)  #畫矩形框
                class_name = L[cls_id]
                cv2.putText(frame, '{:s} | {:.3f}'.format(class_name, score), (xmin, ymin - 2,), cv2.FONT_HERSHEY_COMPLEX, 0.5, (0, 255, 0), 1)  #寫文字
        cv2.imshow("video", frame)
        c = cv2.waitKey(50)
        if c == 27: #esc退出
            break


#Load_video_show()
demo()