LinkToPast1990

Hi, @dorarad Could you kindly tell me when will you release the features for the ground-truth bounding boxes

Processed SOP and In-Shop dataset

1

Hi, @bnu-wangxun Thanks for your great work. Could you kindly provide the processed [SOP](https://cvgl.stanford.edu/projects/lifted_struct/) and [In-Shop](http://mmlab.ie.cuhk.edu.hk/projects/DeepFashion/InShopRetrieval.html), or the pre-processing codes?

Getting Epoch-600 0.7719 0.8504 0.9052 0.9451 0.9701 0.9847 with Binomial on Cars196

1

Hi, @bnu-wangxun I want to reproduce the result of Binomial Loss, so I modified the run_train_00.sh slightly: ``` DATA=car ... LOSS=Binomial ... DIM=64 ``` And I downloaded the bninception from...

oov of gqa answer (gqa/trainval_ans2label.json only has 1842 answers)

4

Hi, @airsplay The GQA paper mentions that there are 1878 possible answers, but I could only collect 1842 ones on the train and valid data. How to deal with the...

About ”tile bbox“

2

Hi, @ronghanghu , is there any reference paper about this operation? ``` # add bounding boxes to the object features # tile bbox to roughly match the l2 norm of...