Skip to content
pytorch implementation for paper, towards realistic predictors
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
models
.DS_Store
README.md
datasets.py
extra_setting.py
realistic_datasets.py
train_HPnet_imagenet_res_res.py
train_HPnet_imagenet_vgg_vgg.py
train_HPnet_indoor_res_res.py
train_HPnet_indoor_vgg_vgg.py
train_rp_imagenet_res_res.py
train_rp_imagenet_vgg_vgg.py
train_rp_indoor_vgg_vgg.py

README.md

towards-realistic-predictors

This repo constains the pytorch implementation for the paper Towards Realistic Predictors on ECCV2018.

requirements

  • python 3.6
  • pytorch = 0.4
  • other common modules

Usage

Since our code is not an end-to-end trainable model (for a improved new version, please turn to an end to end implementation for realistic predictors), please run

train_HPnet_dataset_net_net.py

to get the hardness predictor after replacing 'dataset' and 'net' with 'imagenet' or 'indoor' and 'res' or 'vgg'.

Then run

train_rp_dataset_net_net.py

to get realistic predictors.

We used the custom data load form, so before training, please first make a train and test sample list. Each item is formed as

imagepath target index

Contact

For questions, feel free to reach

Pei Wang: peiwang062@gmail.com
You can’t perform that action at this time.