jiawenhao
jiawenhao
the same question~~~~thanks
> annotations style in COCO must be binary. So, what is your json file generated by, is it based on the whole person or the body part? Thanks. hello~~~do you...
> I just added the path of the annotation to the json file in coco format.(e.g. > annotation_info = { > "id": annotation_id, > "image_id": image_id, > "bbox": bounding_box.tolist(), >...
> Yes, the json file actually contains label annotation's path and used[ pycococreator](https://github.com/waspinator/pycococreator). ok ~~~~~~thanks a lot~~~~~~~~~~~i will try ~~~
@ziqiangchen 我碰到了跟您一模一样的问题。。。维度不匹配，想问一下您是怎么解决的呢？ I0809 07:53:56.976981 4128 net.cpp:131] Top shape: 523 512 2 16 (8568832) I0809 07:53:56.977007 4128 net.cpp:139] Memory required for data: 3418995348 I0809 07:53:56.977032 4128 layer_factory.hpp:77] Creating layer reshape
@ziqiangchen 嗯呢 多谢回复了哈 刚刚看了一下log 之前是523×512×4×16然后经过了一个卷积层 尺寸为3*3 步长为1，但是默认的参数给的pad_w为1，而不是pad为1，然后就导致了宽度没变然后高度变为原来的一半了， 也就是523×512×2×16。刚刚把pad_w改成了pad就可以啦 然后也改了后边对应的尺寸。现在正在跑~训练数据集也都是按examples/crnn下的脚本生成的。。 都是默认的目前已经迭代了3600次ctc的loss发现已经降了一些，但是为啥accuracy一直是0呢。。。
~.~ waiting soluting....
@leonmakise hello,i met the same error,when i tried distributed training.. ...i see your changes,but i can not understand what it means > It shall be self.model.cuda() ori code: `self.model =...
噢，刚刚看到楼主的另外一个repository里面提到生成dll的方法了。抱歉，是我没看仔细。。但是感觉还是放到一个工程里面更合适，或者这个工程里的readme.md里面有提到关于dll的事情等。
pytorch version should be 1.4.0