FPN在faster_rcnn中實現細節代碼說明

代碼參考自：https://github.com/DetectionTeamUCAS/FPN_Tensorflow

主要分析fpn多層金字塔結構的輸出如何進行預測。

FPN金字塔結構插入在faster_rcnn的特征圖獲取之后，在rpn結構之前。

具體代碼如下所示：

代碼結構追溯至FPN部分：

train.py(line 46 :build_whole_detection_network函數）

　　　　build_whole_network(line 372:  build_whole_detection_network函數）

　　　　按照注釋分別查看7個步驟：

　　　　1. build base network
　　　　
　　　　2. build rpn
　
　　　　3. generate_anchors

　　　　4. postprocess rpn proposals. such as: decode, clip, NMS(所得第一次框處理)

　　　　5. build Fast-RCNN（5，roipooling  6,inference rois to obtain fc   7,cls reg）

　　　　6. postprocess_fastrcnn(最后框處理)
　　　　FPN部分在build base network中，得到的plist即為金字塔特征圖的集合
　　　　　　　　build_base_network一步一步回溯找到原函數resnet_base(fpn操作在這里，如下代碼)

def resnet_base(img_batch, scope_name, is_training=True):
    '''
    this code is derived from light-head rcnn.
    https://github.com/zengarden/light_head_rcnn

    It is convenient to freeze blocks. So we adapt this mode.
    '''
    if scope_name == 'resnet_v1_50':
        middle_num_units = 6
    elif scope_name == 'resnet_v1_101':
        middle_num_units = 23
    else:
        raise NotImplementedError('We only support resnet_v1_50 or resnet_v1_101. Check your network name....yjr')

    blocks = [resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
              resnet_v1_block('block2', base_depth=128, num_units=4, stride=2),
              resnet_v1_block('block3', base_depth=256, num_units=middle_num_units, stride=2),
              resnet_v1_block('block4', base_depth=512, num_units=3, stride=1)]
    # when use fpn . stride list is [1, 2, 2]

    with slim.arg_scope(resnet_arg_scope(is_training=False)):
        with tf.variable_scope(scope_name, scope_name):
            # Do the first few layers manually, because 'SAME' padding can behave inconsistently
            # for images of different sizes: sometimes 0, sometimes 1
            net = resnet_utils.conv2d_same(
                img_batch, 64, 7, stride=2, scope='conv1')
            net = tf.pad(net, [[0, 0], [1, 1], [1, 1], [0, 0]])
            net = slim.max_pool2d(
                net, [3, 3], stride=2, padding='VALID', scope='pool1')

    not_freezed = [False] * cfgs.FIXED_BLOCKS + (4-cfgs.FIXED_BLOCKS)*[True]
    # Fixed_Blocks can be 1~3

    with slim.arg_scope(resnet_arg_scope(is_training=(is_training and not_freezed[0]))):
        C2, end_points_C2 = resnet_v1.resnet_v1(net,
                                                blocks[0:1],
                                                global_pool=False,
                                                include_root_block=False,
                                                scope=scope_name)

    # C2 = tf.Print(C2, [tf.shape(C2)], summarize=10, message='C2_shape')
    add_heatmap(C2, name='Layer2/C2_heat')

    with slim.arg_scope(resnet_arg_scope(is_training=(is_training and not_freezed[1]))):
        C3, end_points_C3 = resnet_v1.resnet_v1(C2,
                                                blocks[1:2],
                                                global_pool=False,
                                                include_root_block=False,
                                                scope=scope_name)

    # C3 = tf.Print(C3, [tf.shape(C3)], summarize=10, message='C3_shape')
    add_heatmap(C3, name='Layer3/C3_heat')
    with slim.arg_scope(resnet_arg_scope(is_training=(is_training and not_freezed[2]))):
        C4, end_points_C4 = resnet_v1.resnet_v1(C3,
                                                blocks[2:3],
                                                global_pool=False,
                                                include_root_block=False,
                                                scope=scope_name)

    add_heatmap(C4, name='Layer4/C4_heat')

    # C4 = tf.Print(C4, [tf.shape(C4)], summarize=10, message='C4_shape')
    with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
        C5, end_points_C5 = resnet_v1.resnet_v1(C4,
                                                blocks[3:4],
                                                global_pool=False,
                                                include_root_block=False,
                                                scope=scope_name)
    # C5 = tf.Print(C5, [tf.shape(C5)], summarize=10, message='C5_shape')
    add_heatmap(C5, name='Layer5/C5_heat')

    feature_dict = {'C2': end_points_C2['{}/block1/unit_2/bottleneck_v1'.format(scope_name)],
                    'C3': end_points_C3['{}/block2/unit_3/bottleneck_v1'.format(scope_name)],
                    'C4': end_points_C4['{}/block3/unit_{}/bottleneck_v1'.format(scope_name, middle_num_units - 1)],
                    'C5': end_points_C5['{}/block4/unit_3/bottleneck_v1'.format(scope_name)],
                    # 'C5': end_points_C5['{}/block4'.format(scope_name)],
                    }

    # feature_dict = {'C2': C2,
    #                 'C3': C3,
    #                 'C4': C4,
    #                 'C5': C5}

    pyramid_dict = {}
    with tf.variable_scope('build_pyramid'):
        with slim.arg_scope([slim.conv2d], weights_regularizer=slim.l2_regularizer(cfgs.WEIGHT_DECAY),
                            activation_fn=None, normalizer_fn=None):

            P5 = slim.conv2d(C5,
                             num_outputs=256,
                             kernel_size=[1, 1],
                             stride=1, scope='build_P5')
            if "P6" in cfgs.LEVLES:
                P6 = slim.max_pool2d(P5, kernel_size=[1, 1], stride=2, scope='build_P6')
                pyramid_dict['P6'] = P6

            pyramid_dict['P5'] = P5

            for level in range(4, 1, -1):  # build [P4, P3, P2]

                pyramid_dict['P%d' % level] = fusion_two_layer(C_i=feature_dict["C%d" % level],
                                                               P_j=pyramid_dict["P%d" % (level+1)],
                                                               scope='build_P%d' % level)
            for level in range(4, 1, -1):
                pyramid_dict['P%d' % level] = slim.conv2d(pyramid_dict['P%d' % level],
                                                          num_outputs=256, kernel_size=[3, 3], padding="SAME",
                                                          stride=1, scope="fuse_P%d" % level)
    for level in range(5, 1, -1):
        add_heatmap(pyramid_dict['P%d' % level], name='Layer%d/P%d_heat' % (level, level))

    # return [P2, P3, P4, P5, P6]
    print("we are in Pyramid::-======>>>>")
    print(cfgs.LEVLES)
    print("base_anchor_size are: ", cfgs.BASE_ANCHOR_SIZE_LIST)
    print(20 * "__")
    return [pyramid_dict[level_name] for level_name in cfgs.LEVLES]
    # return pyramid_dict  # return the dict. And get each level by key. But ensure the levels are consitant
    # return list rather than dict, to avoid dict is unordered

觀察原特征圖的結構C2,C3,C4,C5,  以及特征金字塔的結構P5,P4,P3,P2，為5層的特征金字塔結構 。

操作如圖：

金字塔結構的總層數為（p5,p6,p4,p3,p2)

P5 = conv2d(C5)                 (因金字塔特征圖每層的構造是 上面一層的2x upsaming 和左邊的1*1conv后的結果相加)

P6 = max_pool(P5)

 

核心的融合部分在下面代碼中顯示：

P4 = C4 + P5

P3 = C3 + P4

P2 = C2 + P3

            for level in range(4, 1, -1):  # build [P4, P3, P2]

                pyramid_dict['P%d' % level] = fusion_two_layer(C_i=feature_dict["C%d" % level],
                                                               P_j=pyramid_dict["P%d" % (level+1)],
                                                               scope='build_P%d' % level)

 

最后的P_LIST共有：　　LEVLES = ['P2', 'P3', 'P4', 'P5', 'P6']層級

 對應每層特征圖設置不同大小的anchors

　　Instead, we assign anchors of a single scale to each level. Formally, we define the anchors to have areas of {32,64,128,256,512} pixels on {P2,P3,P4,P5,P6} respectively.

As in [29] we also use anchors of multiple aspect ratios{1:2, 1:1, 2:1}at each level. So in total there are 15 anchors over the pyramid

后面得到全部金字塔特征圖的roi，下一步是要把roi對應到各自層的特征圖上取roi特征，不同大小的roi對應不同的特征圖，較大的roi對應深層的特征圖，按照公式

 

 確定對應的層，然后送入roi pooling，統一特征圖尺寸。

 

在roipooling之后的處理過程都基本一樣了；

原來是一個map進行predict產生一些proposals；經過處理之后，送入全連接層之后進行cls  and reg;

FPN現在是多個map進行predict產生更多不同尺度(更加魯棒)的proposals，經過處理之后，也是送入全連接層之后進行cls and reg。