How much Position Information Do Convolutional Neural Networks Encode?
Md Amirul Islam*, Sen Jia*, Neil Bruce

Decoding Absolute Position Information

Our study attemts to demystify if a pre-trained model contains absolute position information, the weight of the backbone is freezed. The simple readout is trainable in order to extract position information from the backbone as much as it can. All the model definitions are under the folder "models". The synthetic images(black, white, noise) and the groundtruth(horizontal, vertical) are under the folder "synthetic".

We train the whole system on the DUT-S dataset and validate on the PASCAL-S dataset, they both are originally used for salient object detection. The position information we explored is content-agnostic, so any natural images can be used. You might want to avoid the ImageNet dataset, because the backbone(vgg) is commonly pre-trained on the data. Please run the following commands to train and evaluate the network.

        python train_network.py folder $abs_train_folder $abs_test_filder

BibTeX

If you find this repository useful, please consider giving a star :star: and citation :t-rex:

  @InProceedings{islam2020position,
   title={How much Position Information Do Convolutional Neural Networks Encode?},
   author={Islam, Md Amirul and Jia, Sen and Bruce, Neil},
   booktitle={International Conference on Learning Representations},
   year={2020}
 }