Heatmap

Heatmap approach for Fashion AI keypoint

Preprocessing

1. split train to trainminusval and val

train

1. cd ./experiments/CPM_FPN/
2. python ./train_net

eval

1. cd ./evaluation
2. python ./csv_evaluation_FPN.py

experiments

1. CPM -> 23% on leaderboard
2. CPM_ResNet 17.9% on valset
3. CPM_FPN + data_aug -> 11% on valset, 12% on leaderboard

#---------------------------------------------- Related Papers--------------------------------------------------------------

1. Attentive Fashion Grammar Network for Fashion Landmark Detection and Clothing Category Classification (BIT, UCLA) - CVPR 2018

• Worthy reading

2. Fashion Landmark Detection in the Wild (Sensetime) - ECCV 2016

• Don't waste your time reading this paper, unless you want to learn from scratch

3. A Coarse-Fine Network for Keypoint Localization (U of Sydeny) - ICCV 2017

• Worthy reading. 非常有意思，文章有两个部分，第一个部分Coarse利用detection的方式检测潜在的目标区域，第二个部分利用softmax的方式分类点。