Li-Chengyang / MSDS-RCNN

''Multispectral Pedestrian Detection via Simultaneous Detection and Segmentation''. BMVC 2018.
MIT License
64 stars 19 forks source link

Iterations required for the pretrained model #10

Open amlandas78 opened 5 years ago

amlandas78 commented 5 years ago

I trained the network for 150000 iterations using the pretrained imagenet models. But I am having a very low mAP which is around 0.0042. Can anyone please help me with the configurations used during training your sanitized model.

jdy160012 commented 5 years ago

do you have the training code ?

jdy160012 commented 4 years ago

Hi , Can I see the training code? thank you

jdy160012 commented 4 years ago

I am recurrenting the result recently, but I only finish the demo., I do not find the training code, could you share yours?