https://github.com/amazon-science/few-shot-object-detection-benchmark

Science Score: 26.0%

This score indicates how likely this project is to be science-related based on various indicators:

○
CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
○
DOI references
○
Academic publication links
○
Academic email domains
○
Institutional organization owner
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (12.2%) to scientific vocabulary

Last synced: 9 months ago · JSON representation

Repository

Basic Info

Host: GitHub
Owner: amazon-science
License: apache-2.0
Language: Python
Default Branch: main
Size: 48.8 MB

Statistics

Stars: 11
Watchers: 8
Forks: 1
Open Issues: 3
Releases: 0

Created almost 4 years ago · Last pushed over 3 years ago

Metadata Files

Readme Contributing License Code of conduct

Rethinking Few-Shot Object Detection on a Multi-Domain Benchmark

This is the PyTorch implementation of Our ECCV paper. It proposes a benchmark for multi-domain few-shot object detection.

Datasets

If you use the code/model/results of this repository please cite:

@inproceedings{lee2022mofsod,
  author    = {Kibok Lee and
               Hao Yang and
               Satyaki Chakraborty and
               Zhaowei Cai and
               Gurumurthy Swaminathan and
               Avinash Ravichandran and
               Onkar Dabeer},
  title     = {Rethinking Few-Shot Object Detection on a Multi-Domain Benchmark},
  booktitle = {ECCV}},
  year      = {2022},
}

Please also cite all the datasets used in the paper. If you want to run the baseline experiments, please cite the corresponding paper as well.

Installation

In this repo, we mainly provide the scripts needed to run MoFSOD benchmarks on top of several few-shot/general detection methods. To run baselines in the paper, you need to checkout these repos: DeFRCN, FSCE, FsDet ,UniDet and Detic. Please follow each repo's instructions to install.

All of the above codebasese are based on Detectron2. Due to compatibility issues, we use Detectron2-v0.3 for FSCE and FsDet, and Detectron2-v0.5 for DeFRCN, UniDet and Detic. Our suggestions is to build two docker images, one with Detectron2-v0.3 and the other with Detectron2-v0.5 and use separate dockers for different experiments.

Dataset Preparation

Please refer the readme under datasets folder to setup all 10 datasets in the benchmark.

We also provide the converted OpenImages style annotations for easier setup. You still need to follow the readme to download and setup JPEGImages folder, then extract and write the datasets folder with the datasets tar file.

Few-shot Sampling

To create the sampling we used in the paper, we prepare a script with fixed random seeds. After setting up the datasets folder, just run

python generate_oi_annotations.py

to create the few-shot sampled datasets. We also provide the exact sampling we used in the paper.

Running on each baseline

Due to the testing needed, we will release this part in a few weeks, stay tune! In the mean time, if you want to test the datasets, we add an OpenImages dataset reader to use in Detectron2. You can register the datasets with

from common.openimages_dataset import openimages
dataset = openimages(data_dir='./datasets/', dataset_name=deepfruits, post_fix_key=str(fold))

Owner

Name: Amazon Science
Login: amazon-science
Kind: organization

Website: https://amazon.science
Twitter: AmazonScience
Repositories: 80
Profile: https://github.com/amazon-science

GitHub Events

Total

Last Year

Issues and Pull Requests

Last synced: over 1 year ago

All Time

Total issues: 3
Total pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Total issue authors: 3
Total pull request authors: 0
Average comments per issue: 0.67
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

Past Year

Issues: 0
Pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Issue authors: 0
Pull request authors: 0
Average comments per issue: 0
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Open Source Science