BEV-CV - (Processing codebase ASAP)

This repository is the official implementation of BEV-CV: Birds-Eye-View Transform for Cross-View Geo-Localisation, IROS 2024.

Requirements

To install requirements:

pip install -r requirements.txt

Datasets - CVUSA & CVACT

We use two existing dataset to do the experiments

CVUSA: sampled across the US, ground-level panoramas and corresponding satellite images. The dataset can be accessed from https://github.com/viibridges/crossnet

CVACT: sampled across Australia, ground-level panoramas and corresponding satellite images. The dataset can be accessed from https://github.com/Liumouliu/OriCNN

Data Preparation

We use limited-FOV panorama crops - emulating a vehicle mounted front-facing monocular camera. To achieve this - the data can be preprocessed before evaluation by running the script:

python run.....

Training

To train the model(s) in the paper, run this command:

python train.py --input-data <path_to_data> --alpha 10 --beta 20

Evaluation

To evaluate my model on ImageNet, run:

python eval.py --model-file mymodel.pth --benchmark imagenet

Pre-trained Models

You can download pretrained models here:

• My awesome model trained on ImageNet using parameters x,y,z.

Results

Our model achieves the following performance on :

Image Retrieval on CVUSA & CVACT

Model name	Top 1 Accuracy	Top 5 Accuracy
My awesome model	85%	95%

Citation

This work is published in IROS 2024. If you are interested in our work and/or use our code, please include the following citation in your work:

  @INPROCEEDINGS{bevcv,
    author={Shore, Tavis and Hadfield, Simon and Mendez, Oscar },
    booktitle={2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
    title={BEV-CV: Birds-Eye-View Transform for Cross-View Geo-Localisation}, 
    year={2024},
    pages={11047-11054},
  }

Contributing

📋 Pick a licence and describe how to contribute to your code repository.