Papers-with-Codes of Few-Shot Object Detection

1. Few Shot Object Detection Preliminaries

• K-shot N-way Object Detection

  • K means the number of the objects for each class
  • N means the number of classes for few shot detection

• DataSet Split

  • Base ClassSet 𝐶_𝑏 , Base Dataset 𝐷_𝑏 contains {(𝑥_𝑖, 𝑦_𝑖)} about abundant images and annotations
  • Novel ClassSet 𝐶_𝑛 , Novel Dataset 𝐷_𝑛 contains {(𝑥_𝑖, 𝑦_𝑖)} about few images and annotations
  • 𝐶_𝑏 ∩ 𝐶_𝑛=∅ , 𝐶_𝑏 ∪ 𝐶_𝑛=𝐶_𝑡𝑜𝑡𝑎𝑙
  • COCO (Base : Novel = 60:20)、 PASCAL VOC (Base : Novel = 15:5)

• Training(Two phases)

  • Training on Base dataset
  • Fine-tuning on Novel and base dataset with few objects

• Method

  • Meta-Learning Based

  • Transfer-Learning Based

2. Recent work

2018

• (AAAI 2018) LSTD: A Low-Shot Transfer Detector for Object Detection [Paper]

2019

• (ICCV 2019) Few-shot Object Detection via Feature Reweighting [Paper] [Code]
• (ICCV 2019) Meta-Learning to Detect Rare Objects [Paper]
• (ICCV 2019) Meta R-CNN : Towards General Solver for Instance-level Low-shot Learning [Paper] [Code]
• (CVPR 2020) RepMet: Representative-based metric learning for classification and few-shot object detection [Paper] [Code]

2020

• (ICML 2020) Frustratingly Simple Few-Shot Object Detection [Paper] [Code]
• (ECCV 2020) Multi-Scale Positive Sample Refinement for Few-Shot Object Detection [Paper] [Code]
• (ECCV 2020) Few-Shot Object Detection and Viewpoint Estimation for Objects in the Wild [Paper] [Code]
• (CVPR 2020) Few-Shot Object Detection with Attention-RPN and Multi-Relation Detector [Paper] [Code]
• (AAAI 2020) Context-Transformer: Tackling Object Confusion for Few-Shot Detection [Paper] [Code]
• (CVPR 2020) Incremental Few-Shot Object Detection [Paper]

2021

• (arxiv) Meta-DETR: Few-Shot Object Detection via Unified Image-Level Meta-Learning [Paper] [Code]

• (CVPR 2021) Semantic Relation Reasoning for Shot-Stable Few-Shot Object Detection [Paper]

• (CVPR 2021) Few-Shot Object Detection via Classification Refinement and Distractor Retreatment [Paper]

• (CVPR 2021) Generalized Few-Shot Object Detection without Forgetting [Paper] [Code]

• (CVPR 2021) FSCE: Few-Shot Object Detection via Contrastive Proposal Encoding [Paper] [Code]

• (CVPR 2021) Dense Relation Distillation with Context-aware Aggregation for Few-Shot Object Detection [Paper] [Code]

• (CVPR 2021) Hallucination Improves Few-Shot Object Detection [Paper] [Code]

• (ICCV 2021) DeFRCN: Decoupled Faster R-CNN for Few-Shot Object Detection [Paper] [Code]

• (AAAI 2021) StarNet: towards Weakly Supervised Few-Shot Object Detection [Paper]

• (CVPR 2021)Beyond Max-Margin: Class Margin Equilibrium for Few-shot Object Detection [Paper] [Code]

• (ICCV 2021) Universal-Prototype Augmentation for Few-Shot Object Detection [Paper] [Code]

• (MM 2021) Dual-awareness Attention for Few-Shot Object Detection [Paper]

• (arxiv) Meta Faster R-CNN: Towards Accurate Few-Shot Object Detection with Attentive Feature Alignment [Paper]

• (arxiv) Class-Incremental Few-Shot Object Detection [Paper]

• (arxiv) Dynamic Relevance Learning for Few-Shot Object Detection [Paper] [Code]

• (NeurIPS 2021) Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental Object Detection [Paper] [Code]

• (NeurIPS 2021 Workshop) Meta Guided Metric Learner for Overcoming Class Confusion in Few-Shot Road Object Detection [Paper]

• (NeurIPS 2021) Few-Shot Object Detection via Association and DIscrimination [Paper] [Code]

• (CVPR 2021) Transformation invariant few- shot object detection [Paper]

• (ICCV 2021) Query Adaptive Few-Shot Object Detection with Heterogeneous Graph Convolutional Networks [Paper]

2022

• (CVPR 2022) Label, Verify, Correct: A Simple Few Shot Object Detection Method [Paper]