is top2down or botton2top? i find the model input is raw image not crop image

svip-lab / HRNet-for-Fashion-Landmark-Estimation.PyTorch

[DeepFashion2 Challenge] Fashion Landmark Estimation with HRNet

MIT License

132 stars 22 forks source link

is top2down or botton2top? i find the model input is raw image not crop image #4

Closed hyaihjq closed 4 years ago

ShenhanQian commented 4 years ago

This is a top-down model. You can choose to feed in the ground-truth bounding boxes or predicted bounding boxes by setting the USE_GT_BBOX option. Then the program will crop the input image into patches for each instance before feeding them into the HRNet.

hyaihjq commented 4 years ago

thanks