beacandler / r2cnn Goto Github PK

之前自己跑其他模型，获得一个icdar2015格式的结果，想要用eval_icdar15.py进行测试，但是报错了

你好，你有没有加入FPN

make:进入目录'/home/xuy/code/R2CNN/src/lib/fastnms'
make: “adaptor.so”是最新的。
make:离开目录“/home/xuy/code/R2CNN/src/lib/fastnms”
WARNING: Logging before InitGoogleLogging() is written to STDERR
W0407 13:56:59.493078 26902 _caffe.cpp:139] DEPRECATION WARNING - deprecated use of Python interface
W0407 13:56:59.493132 26902 _caffe.cpp:140] Use this instead (with the named "weights" parameter):
W0407 13:56:59.493141 26902 _caffe.cpp:142] Net('/home/xuy/code/R2CNN/model/prototxt/test/TextBoxes-v3.prototxt', 1, weights='/home/xuy/code/R2CNN/model/caffemodel/TextBoxes-v2_iter_12w.caffemodel')
[libprotobuf ERROR google/protobuf/text_format.cc:274] Error parsing text-format caffe.NetParameter: 483:21: Message type "caffe.LayerParameter" has no field named "roi_pooling_param".
F0407 13:56:59.494863 26902 upgrade_proto.cpp:88] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: /home/xuy/code/R2CNN/model/prototxt/test/TextBoxes-v3.prototxt
*** Check failure stack trace: ***
Aborted (core dumped)

请问一下作者，我下载了好多icdar2015，请问一下应该用哪个测试本repo呢？
第一个

第二个

Thank your for sharing your code.I did not find the "logs" file in your project,but your "code structor" contains a "logs" file.Maybe i make something wrong? Looking forward to your reply.

您好，看到您说 or you can load cached roidb data of this project, and place it to src/data/ ， [](可是我下载代码看了之后发现https://github.com/beacandler/R2CNN/blob/19d23828885010de4f4411c4ccd01a3032416c4e/src/lib/roi_data_layer/roidb.py#L27， 您的roidb数据的len是15000多个，可是依照我们的icdar的dataset， len(imdb.image_index)只有2000，所以您的大部分数据无法进行roidb[i]['image'] = imdb.image_path_at(i) roidb[i]['width'] = sizes[i][0] roidb[i]['height'] = sizes[i][1] 这些操作，因为i只能到2000，所以能不能麻烦您提供一下dataset, 需要里边的image信息才能train起来，或者您生成这个merged dataset的代码

icdar.py当中看到了两个txt文件，在py-faster rcnn当中会在ImageSets/Main生成trainval.txt，这个trainval.txt对应图片当中的哪个文件呢？merge_train.txt里面的内容是什么呢？

你好，我在test输出数据的时候print了predbbox和score，发现bbox的数据集中在第二个bbox中：
-> print(scores[index])
(Pdb) r
[9.9988842e-01 6.6253820e-06 5.9292459e-05 4.0550061e-05 5.0738868e-06]
[ 0. 0. 0. 0. 0. 0.
0. 0. 43.752 96.91591 482.25037 96.84546
482.31024 469.7978 43.811916 469.86823 0. 0.
0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.
0. 0. 0. 0. ]
于是我就在猜想您的代码原来是写来检测text的，即检测单类别，那我如果用来检测多类别是否有哪里需要改动？