search

yuhuayc / da-faster-rcnn

An implementation of our CVPR 2018 work 'Domain Adaptive Faster R-CNN for Object Detection in the Wild'
Other
336 stars 70 forks source link

question about anchor scales #20

Open xiong233 opened 5 years ago

xiong233 commented 5 years ago

Hello, I am running the original baseline from KITTI to cityscape. When I use ANCHOR_SCALES=[8,16,32], I get similar results in your paper, but when I adjust ANCHOR_SCALES to [4, 8, 16,32], I got a very high result about 42% on the baseline, much higher than the results of your paper. It seems that domain adaptation on this dataset doesn't seem to work, but it's also possible that something went wrong when I implemented it. Could you set ANCHOR_SCALES to [4, 8,16,32] and run it? Thank You!

yuhuayc commented 5 years ago

That's very interesting! thanks for the information. I will try to do some experiments and let you know the outcome!

MaddyThakker commented 4 years ago

@yuhuayc any updates on this?

MaddyThakker commented 4 years ago

@xiong233 while replicating the results, in the source dataset, did you use only "car" class bounding boxes for training?