PyTorch implementation of ''Background Activation Suppression for Weakly Supervised Object Localization''.

Related tags

GeolocationBAS
Overview

Background Activation Suppression for Weakly Supervised Object Localization

PyTorch implementation of ''Background Activation Suppression for Weakly Supervised Object Localization''. This repository contains PyTorch training code, inference code and pretrained models.

📋 Table of content

  1. 📎 Paper Link
  2. 💡 Abstract
  3. Motivation
  4. 📖 Method
  5. 📃 Requirements
  6. ✏️ Usage
    1. Start
    2. Download Datasets
    3. Training
    4. Inference
  7. 📊 Experimental Results
  8. ✉️ Statement
  9. 🔍 Citation

📎 Paper Link

Background Activation Suppression for Weakly Supervised Object Localization (link)

  • Authors: Pingyu Wu*, Wei Zhai*, Yang Cao
  • Institution: University of Science and Technology of China (USTC)

💡 Abstract

Weakly supervised object localization (WSOL) aims to localize the object region using only image-level labels as supervision. Recently a new paradigm has emerged by generating a foreground prediction map (FPM) to achieve the localization task. Existing FPM-based methods use cross-entropy (CE) to evaluate the foreground prediction map and to guide the learning of generator. We argue for using activation value to achieve more efficient learning. It is based on the experimental observation that, for a trained network, CE converges to zero when the foreground mask covers only part of the object region. While activation value increases until the mask expands to the object boundary, which indicates that more object areas can be learned by using activation value. In this paper, we propose a Background Activation Suppression (BAS) method. Specifically, an Activation Map Constraint module (AMC) is designed to facilitate the learning of generator by suppressing the background activation values. Meanwhile, by using the foreground region guidance and the area constraint, BAS can learn the whole region of the object. Furthermore, in the inference phase, we consider the prediction maps of different categories together to obtain the final localization results. Extensive experiments show that BAS achieves significant and consistent improvement over the baseline methods on the CUB-200-2011 and ILSVRC datasets.

Motivation


Motivation. (A) The entroy value of CE loss $w.r.t$ foreground mask and foreground activation value $w.r.t$ foreground mask. To illustrate the generality of this phenomenon, more examples are shown in the subfigure on the right. (B) Experimental procedure and related definitions. Implementation details of the experiment and further results are available in the Supplementary Material.

Exploratory Experiment

We introduce the implementation of the experiment, as shown in Fig. \ref{Exploratory Experiment} (A). For a given GT binary mask, the activation value (Activation) and cross-entropy (Entropy) corresponding to this mask are generated by masking the feature map. We erode and dilate the ground-truth mask with a convolution of kernel size $5n \times 5n$, obtain foreground masks with different area sizes by changing the value of $n$, and plot the activation value versus cross-entropy with the area as the horizontal axis, as shown in Fig. \ref{Exploratory Experiment} (B). By inverting the foreground mask, the corresponding background activation values for the foreground mask area are generated in the same way. In Fig. \ref{Exploratory Experiment} (C), we show the curves of entropy, foreground activation, and background activation with mask area. It can be noticed that both background activation and foreground activation values have a higher correlation with the mask compared to the entropy. We show more examples in the Supplementary Material.


Exploratory Experiment. Examples about the entroy value of CE loss $w.r.t$ foreground mask and foreground activation value $w.r.t$ foreground mask.

📖 Method


The architecture of the proposed BAS. In the training phase, the class-specific foreground prediction map $F^{fg}$ and the coupled background prediction map $F^{bg}$ are obtained by the generator, and then fed into the activation map constraint module together with the feature map $F$. In the inference phase, we utilize Top-k to generate the final localization map.

📃 Requirements

  • python 3.6.10
  • torch 1.4.0
  • torchvision 0.5.0
  • opencv 4.5.3

✏️ Usage

Start

git clone https://github.com/wpy1999/BAS.git
cd BAS

Download Datasets

Training

We will release our training code upon acceptance.

Inference

To test the CUB models, you can download the trained models from [ Google Drive (VGG16) ], [ Google Drive (Mobilenetv1) ], [ Google Drive (ResNet50) ], [ Google Drive (Inceptionv3) ], then run BAS_inference.py:

cd CUB
python BAS_inference.py --arch ${Backbone}

To test the ILSVRC models, you can download the trained models from [ Google Drive (VGG16) ], [ Google Drive (Mobilenetv1) ], [ Google Drive (ResNet50) ], [ Google Drive (Inceptionv3) ], then run BAS_inference.py:

cd ILSVRC
python BAS_inference.py --arch ${Backbone}

📊 Experimental Results



✉️ Statement

This project is for research purpose only, please contact us for the licence of commercial use. For any other questions please contact [email protected] or [email protected].

🔍 Citation

@inproceedings{BAS,
  title={Background Activation Suppression for Weakly Supervised Object Localization},
  author={Pingyu Wu and Wei Zhai and Yang Cao},
  journal={arXiv preprint arXiv:2112.00580},
  year={2021}
}
Search and download Copernicus Sentinel satellite images

sentinelsat Sentinelsat makes searching, downloading and retrieving the metadata of Sentinel satellite images from the Copernicus Open Access Hub easy

837 Dec 28, 2022
GeoIP Legacy Python API

MaxMind GeoIP Legacy Python Extension API Requirements Python 2.5+ or 3.3+ GeoIP Legacy C Library 1.4.7 or greater Installation With pip: $ pip instal

MaxMind 230 Nov 10, 2022
WIP: extracting Geometry utilities from datacube-core

odc.geo This is still work in progress. This repository contains geometry related code extracted from Open Datacube. For details and motivation see OD

Open Data Cube 34 Jan 09, 2023
This repository contains the scripts to derivate the ENU and ECEF coordinates from the longitude, latitude, and altitude values encoded in the NAD83 coordinates.

This repository contains the scripts to derivate the ENU and ECEF coordinates from the longitude, latitude, and altitude values encoded in the NAD83 coordinates.

Luigi Cruz 1 Feb 07, 2022
Centroids as a Service

Centroids! This application reads a valid geojson FeatureCollection and returns a valid geojson FeatureColleciton of centroids. In the output: All pro

Lyzi Diamond 20 Aug 29, 2021
A modern, geometric typeface by @chrismsimpson (last commit @ 85fa625 Jun 9, 2020 before deletion)

Metropolis A modern, geometric typeface. Influenced by other popular geometric, minimalist sans-serif typefaces of the new millenium. Designed for opt

Darius 183 Dec 25, 2022
Satellite imagery for dummies.

felicette Satellite imagery for dummies. What can you do with this tool? TL;DR: Generate JPEG earth imagery from coordinates/location name with public

Shivashis Padhi 1.8k Jan 03, 2023
A ninja python package that unifies the Google Earth Engine ecosystem.

A Python package that unifies the Google Earth Engine ecosystem. EarthEngine.jl | rgee | rgee+ | eemont GitHub: https://github.com/r-earthengine/ee_ex

47 Dec 27, 2022
Simple, concise geographical visualization in Python

Geographic visualizations for HoloViews. Build Status Coverage Latest dev release Latest release Docs What is it? GeoViews is a Python library that ma

HoloViz 445 Jan 02, 2023
Yet Another Time Series Model

Yet Another Timeseries Model (YATSM) master v0.6.x-maintenance Build Coverage Docs DOI | About Yet Another Timeseries Model (YATSM) is a Python packag

Chris Holden 60 Sep 13, 2022
Imports VZD (Latvian State Land Service) open data into postgis enabled database

Python script main.py downloads and imports Latvian addresses into PostgreSQL database. Data contains parishes, counties, cities, towns, and streets.

Kaspars Foigts 7 Oct 26, 2022
Introduction to Geospatial Analysis in Python

Introduction to Geospatial Analysis in Python This repository is in support of a talk on geospatial data. Data To recreate all of the examples, the da

Dillon Gardner 6 Oct 19, 2022
Open Data Cube analyses continental scale Earth Observation data through time

Open Data Cube Core Overview The Open Data Cube Core provides an integrated gridded data analysis environment for decades of analysis ready earth obse

Open Data Cube 410 Dec 13, 2022
Open GeoJSON data on geojson.io

geojsonio.py Open GeoJSON data on geojson.io from Python. geojsonio.py also contains a command line utility that is a Python port of geojsonio-cli. Us

Jacob Wasserman 114 Dec 21, 2022
Blender addons to make the bridge between Blender and geographic data

Blender GIS Blender minimal version : 2.8 Mac users warning : currently the addon does not work on Mac with Blender 2.80 to 2.82. Please do not report

5.9k Jan 02, 2023
Replace MSFS2020's bing map to google map

English verison here 中文 免责声明 本教程提到的方法仅用于研究和学习用途。我不对使用、拓展该教程及方法所造成的任何法律责任和损失负责。 背景 微软模拟飞行2020的地景使用了Bing的卫星地图,然而卫星地图比较老旧,很多地区都是几年前的图设置直接是没有的。这种现象在全球不同地区

hesicong 272 Dec 24, 2022
How to use COG's (Cloud optimized GeoTIFFs) with Rasterio

How to use COG's (Cloud optimized GeoTIFFs) with Rasterio According to Cogeo.org: A Cloud Opdtimized GeoTIFF (COG) is a regular GeoTIFF file, aimed at

Marvin Gabler 8 Jul 29, 2022
This app displays interesting statistical weather records and trends which can be used in climate related research including study of global warming.

This app displays interesting statistical weather records and trends which can be used in climate related research including study of global warming.

0 Dec 27, 2021
ProjPicker (projection picker) is a Python module that allows the user to select all coordinate reference systems (CRSs)

ProjPicker ProjPicker (projection picker) is a Python module that allows the user to select all coordinate reference systems (CRSs) whose extent compl

Huidae Cho 4 Feb 06, 2022
A service to auto provision devices in Aruba Central based on the Geo-IP location

Location Based Provisioning Service for Aruba Central A service to auto provision devices in Aruba Central based on the Geo-IP location Geo-IP auto pr

Will Smith 3 Mar 22, 2022