suyukun666 / S2CNet

Official PyTorch implementation of the “Spatial-Semantic Collaborative Cropping for User Generated Content”. (AAAI24)
40 stars 4 forks source link

where is the rcnn_bboxes come from? #5

Open ppdk-data opened 8 months ago

ppdk-data commented 8 months ago

where is the rcnn_bboxes come from?

suyukun666 commented 8 months ago

sry for the late reply. As mentioned in the paper, the boxes are generated by faster-rnn

ppdk-data commented 8 months ago

sry for the late reply. As mentioned in the paper, the boxes are generated by faster-rnn

Thanks for your reply. So the implement of faster-rnn and the procdure of generating rcnn_bboxes is not in this codes?

suyukun666 commented 8 months ago

yes,you can refer to the faster rcnn link we provide at the end of our github 发自我的 iPhone

在 2024年3月12日,22:12,ppdk-data @.***> 写道:



sry for the late reply. As mentioned in the paper, the boxes are generated by faster-rnn

Thanks for your reply. So the implement of faster-rnn and the procdure of generating rcnn_bboxes is not in this codes?

— Reply to this email directly, view it on GitHubhttps://github.com/suyukun666/S2CNet/issues/5#issuecomment-1991746855, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AH2A3FLLG6FRO3QTDM6LYKTYX4ENBAVCNFSM6AAAAABERV4HDCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOJRG42DMOBVGU. You are receiving this because you commented.Message ID: @.***>

dongdk commented 2 months ago

just simply use the faster-rcnn output results?

dongdk commented 2 months ago

hi @suyukun666 , i find that s2cnet not just simplifies to use the output from faster-rcnn-vg, it looks like that it selects some bboxes to train or validate, e.g. this pic is visualized provided by the repo image and this pic is visualized by faster-rcnn-vg image

can u explain how to select the outputs from faster-rcnn-vg? thanks.

dongdk commented 2 months ago

oh get it, just sort the scores of the final ouputs to select top-k bbox