Centernet: “RuntimeError: The size of tensor a (38) must match the size of tensor b (37) at non-singleton dimension 2 ”

Created on 23 Jan 2020  ·  3Comments  ·  Source: xingyizhou/CenterNet

why do I have this problem?

picture wwetnaroh

Most helpful comment

The input resolution needs to be divisible by 32 for DLA/ resnets and 128 for HourglassNet.

picture xingyizhou on 27 Jan 2020
👍3

All 3 comments

The input resolution needs to be divisible by 32 for DLA/ resnets and 128 for HourglassNet.

xingyizhou picture xingyizhou on 27 Jan 2020
👍3

how to solve this？
i have the same problem.
inputsize300*300

czy112 picture czy112 on 18 Apr 2020

I get the same error when I try to run centernet with visdrone2019 dataset. The input resolution is already divisible by 32. It seems like I should change something in src/lib/models/networks/pose_dla_dcn.py.

cnr0724 picture cnr0724 on 10 Mar 2021
Was this page helpful?
0 / 5 - 0 ratings

Related issues

Alexyitx picture Alexyitx  ·  6Comments
lih627 picture lih627  ·  7Comments
huang229 picture huang229  ·  4Comments
bhack picture bhack  ·  6Comments
li980432131 picture li980432131  ·  6Comments