andeyeluguo

Results 21 comments of andeyeluguo

Build Fail with WITH_PYTHON_LAYER := 1 in Makefile.config

brother yihui's solution can sovle my problem ,thank you.

Build Fail with WITH_PYTHON_LAYER := 1 in Makefile.config

I solve it by add boost_regex to LIBRARIES in Makefile just like this 263 LIBRARIES += boost_thread stdc++ boost_regex maybe you miss the boost_regex

Build Fail with WITH_PYTHON_LAYER := 1 in Makefile.config

then I have no idea,sorry

where is the absval_layer?

also, why don't use the relu functioin to replace the tanh?

Not getting a good accuracy

I only get 43% val acc when the epoch is 400000, I use your solver.prototxt and change the deploy.prototxt into train_val.prototxt. Is it not sufficient to train？ or the preprocess...

what's the type of "ConvolutionDepthwise"

I found a implementation of this layer in you annother repository.

How about the inference accuracy

Have you really sovled this problem, someone says it is the resize of darknet is letterbox, but it has no this operation in openvino, I have no idea how to...

How about the inference accuracy

@Jucjiaswiss after use the letter-box resize ,have you solved it ?

Openvino performance a lot worse than darknet model

all people meet this bug and nobody can solve it even the intel coporation by now.

Openvino performance a lot worse than darknet model

it may decreased 20 MAP