Mmdetection: any plans and results for the SE154

Created on 3 Oct 2019  路  2Comments  路  Source: open-mmlab/mmdetection

I experiment the HTC model using the senet154 backbone models from cadene. And the result is only 42.3 which is lower than your resnext101 model which is 43.9. In the experiment, I used the DCN and multi-scale training. So if convenient,could you share your SENet154 pretrained model in your HTC paper?

picture runzeer

Most helpful comment

You may expect it in a month.

picture hellock on 3 Oct 2019
👍4

All 2 comments

You may expect it in a month.

hellock picture hellock on 3 Oct 2019
👍4

Hi锛宎ny plans for the SE154 results?

runzeer picture runzeer on 3 Dec 2019
Was this page helpful?
0 / 5 - 0 ratings

Related issues

hust-kevin picture hust-kevin  路  3Comments
Hemantr05 picture Hemantr05  路  3Comments
namheegordonkim picture namheegordonkim  路  3Comments
henbucuoshanghai picture henbucuoshanghai  路  3Comments
michaelisc picture michaelisc  路  3Comments