开源Keras Faster RCNN 模型介绍

Posted 2021-09-06 ZSYL

开源Keras Faster RCNN 模型介绍

• 1. 环境需求
• 2. keras Faster RCNN代码结构
• 3. FaterRCNN源码解析
• 4. RPN 与 classifier定义
• 5. data_generators.py:传递图像参数，增广配置参数，是否进行图像增广
• 6. 损失计算 losses.py

开源Keras Faster RCNN 模型地址：https://github.com/jinfagang/keras_frcnn

1. 环境需求

• 1、由于该源代码由keras单独库编写所以需要下载keras,必须是2.0.3版本

pip install keras==2.0.3

• 2、该源码读取图片以及处理图片标记图片工具使用opecv需要安装

pip install opencv-python

2. keras Faster RCNN代码结构

源码组成结构：

3. FaterRCNN源码解析

• detector：FasterRCNNDetector目标检测器代码

img_input = Input(shape=input_shape_img)
    roi_input = Input(shape=(None, 4))

    # define the base network (resnet here, can be VGG, Inception, etc)
    shared_layers = nn.nn_base(img_input, trainable=True)

    # define the RPN, built on the base layers
    num_anchors = len(cfg.anchor_box_scales) * len(cfg.anchor_box_ratios)
    rpn = nn.rpn(shared_layers, num_anchors)

    classifier = nn.classifier(shared_layers, roi_input, cfg.num_rois, nb_classes=len(classes_count), trainable=True)

    model_rpn = Model(img_input, rpn[:2])
    model_classifier = Model([img_input, roi_input], classifier)

    # this is a model that holds both the RPN and the classifier, used to load/save weights for the models
    model_all = Model([img_input, roi_input], rpn[:2] + classifier)

    try:
        print('loading weights from {}'.format(cfg.base_net_weights))
        model_rpn.load_weights(cfg.model_path, by_name=True)
        model_classifier.load_weights(cfg.model_path, by_name=True)
    except Exception as e:
        print(e)
        print('Could not load pretrained model weights. Weights can be found in the keras application folder '
              'https://github.com/fchollet/keras/tree/master/keras/applications')

    optimizer = Adam(lr=1e-5)
    optimizer_classifier = Adam(lr=1e-5)
    model_rpn.compile(optimizer=optimizer,
                      loss=[losses_fn.rpn_loss_cls(num_anchors), losses_fn.rpn_loss_regr(num_anchors)])
    model_classifier.compile(optimizer=optimizer_classifier,
                             loss=[losses_fn.class_loss_cls, losses_fn.class_loss_regr(len(classes_count) - 1)],
                             metrics={'dense_class_{}'.format(len(classes_count)): 'accuracy'})
    model_all.compile(optimizer='sgd', loss='mae')

4. RPN 与 classifier定义

• RPN结构

def rpn(base_layers, num_anchors):
    x = Convolution2D(512, (3, 3), padding='same', activation='relu', kernel_initializer='normal', name='rpn_conv1')(
        base_layers)

    x_class = Convolution2D(num_anchors, (1, 1), activation='sigmoid', kernel_initializer='uniform',
                            name='rpn_out_class')(x)
    x_regr = Convolution2D(num_anchors * 4, (1, 1), activation='linear', kernel_initializer='zero',
                           name='rpn_out_regress')(x)

    return [x_class, x_regr, base_layers]

• classifier结构

def classifier(base_layers, input_rois, num_rois, nb_classes=21, trainable=False):
    # compile times on theano tend to be very high, so we use smaller ROI pooling regions to workaround

    if K.backend() == 'tensorflow':
        pooling_regions = 14
        input_shape = (num_rois, 14, 14, 1024)
    elif K.backend() == 'theano':
        pooling_regions = 7
        input_shape = (num_rois, 1024, 7, 7)
	
    # ROI pooling计算定义
    out_roi_pool = RoiPoolingConv(pooling_regions, num_rois)([base_layers, input_rois])
    out = classifier_layers(out_roi_pool, input_shape=input_shape, trainable=True)

    out = TimeDistributed(Flatten())(out)

    # 分类
    out_class = TimeDistributed(Dense(nb_classes, activation='softmax', kernel_initializer='zero'),
                                name='dense_class_{}'.format(nb_classes))(out)
    # note: no regression target for bg class
    # 回归
    out_regr = TimeDistributed(Dense(4 * (nb_classes - 1), activation='linear', kernel_initializer='zero'),
                               name='dense_regress_{}'.format(nb_classes))(out)
    return [out_class, out_regr]

5. data_generators.py:传递图像参数，增广配置参数，是否进行图像增广

• IoU计算：

from __future__ import absolute_import
import numpy as np
import cv2
import random
import copy
from . import data_augment
import threading
import itertools

#并集
def union(au, bu, area_intersection):
    area_a = (au[2] - au[0]) * (au[3] - au[1])
    area_b = (bu[2] - bu[0]) * (bu[3] - bu[1])
    area_union = area_a + area_b - area_intersection
    return area_union

#交集
def intersection(ai, bi):
    x = max(ai[0], bi[0])
    y = max(ai[1], bi[1])
    w = min(ai[2], bi[2]) - x
    h = min(ai[3], bi[3]) - y
    if w < 0 or h < 0:
        return 0
    return w*h

#交并比
def iou(a, b):
    # a and b should be (x1,y1,x2,y2)
    if a[0] >= a[2] or a[1] >= a[3] or b[0] >= b[2] or b[1] >= b[3]:
        return 0.0

    area_i = intersection(a, b)
    area_u = union(a, b, area_i)

    return float(area_i) / float(area_u + 1e-6)

6. 损失计算 losses.py

• rpn的损失回归和分类

def rpn_loss_regr(num_anchors):
	def rpn_loss_regr_fixed_num(y_true, y_pred):
		if K.image_dim_ordering() == 'th':
			x = y_true[:, 4 * num_anchors:, :, :] - y_pred
			x_abs = K.abs(x)
			x_bool = K.less_equal(x_abs, 1.0)
			return lambda_rpn_regr * K.sum(
				y_true[:, :4 * num_anchors, :, :] * (x_bool * (0.5 * x * x) + (1 - x_bool) * (x_abs - 0.5))) / K.sum(epsilon + y_true[:, :4 * num_anchors, :, :])
		else:
			x = y_true[:, :, :, 4 * num_anchors:] - y_pred
			x_abs = K.abs(x)
			x_bool = K.cast(K.less_equal(x_abs, 1.0), tf.float32)

			return lambda_rpn_regr * K.sum(
				y_true[:, :, :, :4 * num_anchors] * (x_bool * (0.5 * x * x) + (1 - x_bool) * (x_abs - 0.5))) / K.sum(epsilon + y_true[:, :, :, :4 * num_anchors])

	return rpn_loss_regr_fixed_num


def rpn_loss_cls(num_anchors):
	def rpn_loss_cls_fixed_num(y_true, y_pred):
		if K.image_dim_ordering() == 'tf':
			return lambda_rpn_class * K.sum(y_true[:, :, :, :num_anchors] * K.binary_crossentropy(y_pred[:, :, :, :], y_true[:, :, :, num_anchors:])) / K.sum(epsilon + y_true[:, :, :, :num_anchors])
		else:
			return lambda_rpn_class * K.sum(y_true[:, :num_anchors, :, :] * K.binary_crossentropy(y_pred[:, :, :, :], y_true[:, num_anchors:, :, :])) / K.sum(epsilon + y_true[:, :num_anchors, :, :])

	return rpn_loss_cls_fixed_num

• fastrcnn的分类和回归

def class_loss_regr(num_classes):
	def class_loss_regr_fixed_num(y_true, y_pred):
		x = y_true[:, :, 4*num_classes:] - y_pred
		x_abs = K.abs(x)
		x_bool = K.cast(K.less_equal(x_abs, 1.0), 'float32')
		return lambda_cls_regr * K.sum(y_true[:, :, :4*num_classes] * (x_bool * (0.5 * x * x) + (1 - x_bool) * (x_abs - 0.5))) / K.sum(epsilon + y_true[:, :, :4*num_classes])
	return class_loss_regr_fixed_num


def class_loss_cls(y_true, y_pred):
	return lambda_cls_class * K.mean(categorical_crossentropy(y_true[0, :, :], y_pred[0, :, :]))

加油!

感谢!

努力!