Faster RCNN generate.py

该模块的功能主要是生成rpn proposals。
+ cv2.imread读取的图像的存储格式为H W K，且三通道的顺序为BGR
+ PIL.Image.open读取图片的存储格式为：W H K ，且三通道顺序为RGB
def imdb_proposals(net, imdb):

def imdb_proposals(net, imdb):    """Generate RPN proposals on all images in an imdb."""    _t = Timer()    imdb_boxes = [[] for _ in xrange(imdb.num_images)]    for i in xrange(imdb.num_images):        # cv2.imread读取的图像的存储格式为H W K，且三通道的顺序为BGR        # PIL.Image.open读取图片的存储格式为：W H K ，且三通道顺序为RGB        im = cv2.imread(imdb.image_path_at(i))        _t.tic()        #调用 im_proposals生成单张图片的rpn proposals， 以及得分        imdb_boxes[i], scores = im_proposals(net, im)        _t.toc()        print 'im_proposals: {:d}/{:d} {:.3f}s' \              .format(i + 1, imdb.num_images, _t.average_time)        if 0:            dets = np.hstack((imdb_boxes[i], scores))            # from IPython import embed; embed()            _vis_proposals(im, dets[:3, :], thresh=0.9)            plt.show()    return imdb_boxes

---

def im_proposals(net, im):该方法中会调用网络的forwad，从而得到想要的boxes和scores

def im_proposals(net, im):    """Generate RPN proposals on a single image."""    blobs = {}    #调用_get_image_blob函数将图像转换为caffe所支持的输入数据结构即四维blob    blobs['data'], blobs['im_info'] = _get_image_blob(im)    # *(blobs['data'].shape 中的 ‘*’涉及到了python中函数的参数收集及逆操作    net.blobs['data'].reshape(*(blobs['data'].shape))    net.blobs['im_info'].reshape(*(blobs['im_info'].shape))    blobs_out = net.forward(            data=blobs['data'].astype(np.float32, copy=False),            im_info=blobs['im_info'].astype(np.float32, copy=False))    #返回im_info这个blob中存储的scale，等价于blobs['im_info'][0][2]    scale = blobs['im_info'][0, 2]    # 通过net的前向传播得到boxes, scores，注意，将boxes返回之前，需要将其缩放会原来的size    # blobs_out['rois'][：]的第一位为类别，后四位才是坐标    boxes = blobs_out['rois'][:, 1:].copy() / scale    scores = blobs_out['scores'].copy()    return boxes, scores

下面重点讲一下blobs_out = net.forward(data=blobs['data'].astype(np.float32, copy=False), im_info=blobs['im_info'].astype(np.float32, copy=False)), 当然纯属个人观点：

def _Net_forward(self, blobs=None, start=None, end=None, **kwargs):    """    Forward pass: prepare inputs and run the net forward.    Parameters    ----------    blobs : list of blobs to return in addition to output blobs.    kwargs : Keys are input blob names and values are blob ndarrays.             For formatting inputs for Caffe, see Net.preprocess().             If None, input is taken from data layers.    start : optional name of layer at which to begin the forward pass    end : optional name of layer at which to finish the forward pass          (inclusive)    Returns    -------    outs : {blob name: blob ndarray} dict.    """    if blobs is None:        blobs = []        # 返回name为start的layer的id，作为start_ind        start_ind = list(self._layer_names).index(start)    else:        start_ind = 0    if end is not None:        end_ind = list(self._layer_names).index(end)        outputs = set([end] + blobs)    else:        end_ind = len(self.layers) - 1        outputs = set(self.outputs + blobs)    if kwargs:        if set(kwargs.keys()) != set(self.inputs):            raise Exception('Input blob arguments do not match net inputs.')        # Set input according to defined shapes and make arrays single and        # C-contiguous as Caffe expects.        # in_为blob name， blob为 blob ndarray        for in_, blob in kwargs.iteritems():            if blob.shape[0] != self.blobs[in_].num:                raise Exception('Input is not batch sized')            self.blobs[in_].data[...] = blob    # 对应_caffe.cpp中的.def("_forward", &Net<Dtype>::ForwardFromTo)，可以猜想应该是调用底层的Net<Dtype>::ForwardFromTo方法，进行前向传播    self._forward(start_ind, end_ind)    # Unpack blobs to extract    #rpn_test.pt所定义的网络的output为：rois blob 和 scores blob 两行log可以说明：    #I0429 03:42:10.559293  9520 net.cpp:270] This network produces output rois    #I0429 03:42:10.559300  9520 net.cpp:270] This network produces output scores    #outputs为以列表，其元素为网络所有输出blob的name    return {out: self.blobs[out].data for out in outputs}

1. net.forwar() 调用的是Pycaffe.py中的_Net_forward函数，代码如上：（关于pycaffe.py —-> Wrap the internal caffe C++ module (_caffe.so) with a clean, Pythonic interface.）
2. 在Pycaffe.py中，方法中带有self参数，个人觉得这应该表示一个Net对象，_Net_forward返回一个字典，{blob name: blob ndarray}，
3. start， end为可选的 layer name，注意是name
4. kwargs : Keys are input *blob names and values are blob ndarrays.*
5. self._layer_names：对应_caffe.cpp中的.add_property("_layer_names", bp::make_function(&Net<Dtype>::layer_names，bp::return_value_policy<bp::copy_const_reference>()))，可以猜想应该是调用底层的Net::layer_names，返回网络中所有层的name
6. self._forward(start_ind, end_ind)：对应_caffe.cpp中的.def("_forward", &Net<Dtype>::ForwardFromTo)，可以猜想应该是调用底层的Net<Dtype>::ForwardFromTo方法，进行前向传播

7. outputs = set(self.outputs + blobs)：对应pycaffe.py中

   @propertydef _Net_outputs(self):return [list(self.blobs.keys())[i] for i in self._outputs]

   a) self._outputs（_caffe.cpp）调用底层的Net::output_blob_indices方法，返回网络所有输出blob的id ;
   b) self.blobs.keys():
   self.blobs对应pycaffe.py中的

   def _Net_blobs(self):"""An OrderedDict (bottom to top, i.e., input to output) of networkblobs indexed by name"""return OrderedDict(zip(self._blob_names, self._blobs))

   返回{blob name : blob ndarray} dict
   所以self.outputs返回的应该是网络输出blob的name

总之， _caffe.cpp 和 pycaffe.py这两个文件要好好研究一下

---

def _get_image_blob(im)

def _get_image_blob(im):    """Converts an image into a network input.也就是将图像转换为caffe所支持的输入数据结构即blob    Arguments:        im (ndarray): a color image in BGR order    Returns:        blob (ndarray): a data blob holding an image pyramid        im_scale_factors (list): list of image scales (relative to im) used            in the image pyramid    """    im_orig = im.astype(np.float32, copy=True)    im_orig -= cfg.PIXEL_MEANS    im_shape = im_orig.shape    im_size_min = np.min(im_shape[0:2])    im_size_max = np.max(im_shape[0:2])    processed_ims = []    assert len(cfg.TEST.SCALES) == 1    target_size = cfg.TEST.SCALES[0]    im_scale = float(target_size) / float(im_size_min)    # Prevent the biggest axis from being more than MAX_SIZE    if np.round(im_scale * im_size_max) > cfg.TEST.MAX_SIZE:        im_scale = float(cfg.TEST.MAX_SIZE) / float(im_size_max)    im = cv2.resize(im_orig, None, None, fx=im_scale, fy=im_scale,                    interpolation=cv2.INTER_LINEAR)    # im_info，一些图像信息：H W scale， 而且数据结构为[[H, W, scale]]    #np.newaxis添加了一个新轴，但是，新坐标轴上没有元素    im_info = np.hstack((im.shape[:2], im_scale))[np.newaxis, :]    processed_ims.append(im)    # Create a blob to hold the input images    #调用blob.py中的im_list_to_blob将图像转换为caffe所支持的数据结构blob，所做的工作就是复制数据，调整通道顺序，im_list_to_blob返回的其实是np.ndarray    blob = im_list_to_blob(processed_ims)    #返回的blob, im_info其实都是np.ndarray    return blob, im_info