You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello, chengyangfu. I'm a graduate student. I want to reproduce your achievement. In this process, I have met some problem, for DSSD300, I wonder all the prediction module consist of conv(256), conv(256), conv(1024)? And the all of the deconvolution is 512 dimension? Original SSD's conv4_3(512), fc7(1024), conv8_2(512), conv9_2(256), conv10_2(256), conv11_2(256) , is all of this layer's deconvolution's dimension is 512?
Thank for your work, look forward your reply.
The text was updated successfully, but these errors were encountered:
Hello, chengyangfu. I'm a graduate student. I want to reproduce your achievement. In this process, I have met some problem, for DSSD300, I wonder all the prediction module consist of conv(256), conv(256), conv(1024)? And the all of the deconvolution is 512 dimension? Original SSD's conv4_3(512), fc7(1024), conv8_2(512), conv9_2(256), conv10_2(256), conv11_2(256) , is all of this layer's deconvolution's dimension is 512?
Thank for your work, look forward your reply.
The text was updated successfully, but these errors were encountered: