Few-Shot Object Detection by Second-Order Pooling

Shan, Zhang; Luo, Dawei; Wang, Lei; Koniusz, Piotr

doi:10.1007/978-3-030-69538-5_23

Cited by 25 publications

(24 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where i, j ∈ {1, • • • , H s }. PNSD [92] embeds Second-Order Pooling (SOP) and Power Normalization (PN) [35] into AttentionRPN. SOP extracts second-order statistics of data features but it introduces a nuisance variability, and PN is added to reduce this nuisance variability.…”

Section: Methods Detection Framework Aggregation Methodsmentioning

confidence: 99%

A Survey of Deep Learning for Low-Shot Object Detection

Huang¹,

Zhang²,

Xue³

et al. 2021

Preprint

View full text Add to dashboard Cite

Object detection is a fundamental task in computer vision and image processing. Current deep learning based object detectors have been highly successful with abundant labeled data. But in real life, it is not guaranteed that each object category has enough labeled samples for training. These large object detectors are easy to overfit when the training data is limited. Therefore, it is necessary to introduce few-shot learning and zeroshot learning into object detection, which can be named low-shot object detection together. Low-Shot Object Detection (LSOD) aims to detect objects from a few or even zero labeled data, which can be categorized into few-shot object detection (FSOD) and zero-shot object detection (ZSD), respectively. This paper conducts a comprehensive survey for deep learning based FSOD and ZSD. First, this survey classifies methods for FSOD and ZSD into different categories and discusses the pros and cons of them. Second, this survey reviews dataset settings and evaluation metrics for FSOD and ZSD, then analyzes the performance of different methods on these benchmarks. Finally, this survey discusses future challenges and promising directions for FSOD and ZSD.

show abstract

Section: Methods Detection Framework Aggregation Methodsmentioning

confidence: 99%

A Survey of Deep Learning for Low-Shot Object Detection

Huang¹,

Zhang²,

Xue³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…DGI [44] uses the block-contrastive strategy [3] by treating negative samples as a difference of instances and a summary of node embeddings for positive samples. Finally, COLES can be extended to other domains/problems e.g., time series/change point detection [9] or few-shot learning [39,54,55].…”

Section: Related Workmentioning

confidence: 99%

Contrastive Laplacian Eigenmaps

Zhu

Sun²,

Koniusz³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Graph contrastive learning attracts/disperses node representations for similar/dissimilar node pairs under some notion of similarity. It may be combined with a low-dimensional embedding of nodes to preserve intrinsic and structural properties of a graph. In this paper, we extend the celebrated Laplacian Eigenmaps with contrastive learning, and call them COntrastive Laplacian EigenmapS (COLES). Starting from a GAN-inspired contrastive formulation, we show that the Jensen-Shannon divergence underlying many contrastive graph embedding models fails under disjoint positive and negative distributions, which may naturally emerge during sampling in the contrastive setting. In contrast, we demonstrate analytically that COLES essentially minimizes a surrogate of Wasserstein distance, which is known to cope well under disjoint distributions. Moreover, we show that the loss of COLES belongs to the family of so-called block-contrastive losses, previously shown to be superior compared to pair-wise losses typically used by contrastive methods. We show on popular benchmarks/backbones that COLES offers favourable accuracy/scalability compared to DeepWalk, GCN, Graph2Gauss, DGI and GRACE baselines.

show abstract

“…Very recent approaches in metalearning like [38] encourages sharing of information between support and query images to enhance class-specific feature sets While CME [15] establishes an equilibrium between class margins to reduce class confusion and demonstrates better generalization to novel classes. A characteristic feature of meta learners is the use of attention mechanisms [6], [39] to identify the most discriminative features for each class. This allows meta learners to retain the knowledge of base classes while generalizing to novel classes.…”

Section: A Few-shot Object Detectionmentioning

confidence: 99%