Home

Awesome

Global Context-Aware Progressive Aggregation Network for Salient Object Detection

This repo. is an implementation of GCPANet , which is accepted for presentation in AAAI 2020.

GCPANet

vis

dependencies

>= Ubuntu 16.04 
>= Python3.5
>= Pytorch 1.0.0
OpenCV-Python

preparation

training

you may revise the TAG and SAVEPATH defined in the train.py. After the preparation, run this command

python3 train.py

make sure that the GPU memory is enough (the original training is conducted on a NVIDIA RTX (24G) card with the batch size of 32).

test

After the preparation, run this commond

 python3 test.py model/model-xxxxx.pt

We provide the trained model file (Google drive), and run this command to check its completeness:

cksum model-100045448.pt 

you will obtain the result 100045448 268562671 model_100045448.pt. The saliency maps are also available (Google drive).

evaluation

We provide the evaluation code in the folder "eval_code" for fair comparisons. You may need to revise the algorithms , data_root, and maps_root defined in the main.m. The saliency maps of the competitors are provided (Google drive).