how about one stage method? about crowddet HOT 7 CLOSED

So far, we haven't tried the method on one-stage method.

from crowddet.

please where should I add the loss if I want to modify it on the one-stage model?

from crowddet.

Maybe you can try to let the anchor boxes predict multiple predictions.

from crowddet.

I'll try to implement a one-stage version based on RetinaNet and update the code if it works.
I think how to combine the Emd Loss and Focal Loss is the key to solve the problem.

from crowddet.

thx, hope your results.

from crowddet.

Applying emd loss to the Retinanet can also have gain in all evaluation indicators.
The result of a simple Retinanet version is as follows, and the code will be released soon.
retina_fpn
mAP:0.8180, mMR:0.5619, JI:0.7312
retina_emd
mAP:0.8303, mMR:0.5537, JI:0.7370

from crowddet.

The new code include the Retinanet implementation.

from crowddet.