Balanced and Hierarchical Relation Learning for One-shot Object Detection

Yang, Hanqing; Cai, Sijia; Sheng, Hualian; Deng, Bing; Huang, Jianqiang; Hua, Xian-Sheng; Tang, Yong

doi:10.1109/cvpr52688.2022.00744

Cited by 10 publications

(15 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Table 2, we compare our method with two‐branch detectors including Meta R‐CNN [29], Attention‐RPN [46], FsDetView [39], Dense Relation Distillation with Context‐aware Aggregation Network [31], CME [14], Transformation Invariant Principle [47], Meta‐DETR [36], Few‐Shot Object Detection with Universal Prototypes [11], Query Adaptive Few‐Shot Object Detection [48], Generate Detectors [49], Meta Faster R‐CNN [50], Intra‐Support Attention Module and the Query‐Support Attention Module [51], CAReD [32] and Kernelized Few‐Shot Object Detection [52], which are mata‐learning‐based methods and single‐branch detectors including TFA [33], MPSR [30], Semantic Relation Reasoning for Shot‐Stable Few‐Shot Object Detection [53], FSCE [13], Cooperating Region Proposal Network’s (CoRPNs) + Hallucination [54], Singular Value Decomposition [55], Few‐Shot Object Detection via Association and Discrimination [56], Decoupled Faster Region based Convolutional Neural Network [57] and TeSNet [35], which are fine‐tuning‐based methods. And it can be seen that our method has a great improvement over other state‐of‐the‐art methods.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Few‐shot object detection via class encoding and multi‐target decoding

Guo

Yang

Wei

et al. 2023

IET Cyber-Syst and Robotics

View full text Add to dashboard Cite

The task of few-shot object detection is to classify and locate objects through a few annotated samples. Although many studies have tried to solve this problem, the results are still not satisfactory. Recent studies have found that the class margin significantly impacts the classification and representation of the targets to be detected. Most methods use the loss function to balance the class margin, but the results show that the loss-based methods only have a tiny improvement on the few-shot object detection problem. In this study, the authors propose a class encoding method based on the transformer to balance the class margin, which can make the model pay more attention to the essential information of the features, thus increasing the recognition ability of the sample. Besides, the authors propose a multi-target decoding method to aggregate RoI vectors generated from multi-target images with multiple support vectors, which can significantly improve the detection ability of the detector for multi-target images. Experiments on Pascal visual object classes (VOC) and Microsoft Common Objects in Context datasets show that our proposed Few-Shot Object Detection via Class Encoding and Multi-Target Decoding significantly improves upon baseline detectors (average accuracy improvement is up to 10.8% on VOC and 2.1% on COCO), achieving competitive performance. In general, we propose a new way to regulate the class margin between support set vectors and a way of feature aggregation for images containing multiple objects and achieve remarkable results. Our method is implemented on mmfewshot, and the code will be available later.

show abstract

Section: Resultsmentioning

confidence: 99%

“…By contrast, humans can easily remember and recognise a kind of object through a few samples. Therefore, researchers have proposed various methods [8][9][10][11] to address the few-shot problem in object detection. These methods can be roughly grouped into two categories: meta-learning-based methods and fine-tuningbased methods.…”

Section: Introductionmentioning

confidence: 99%

Few‐shot object detection via class encoding and multi‐target decoding

Guo

Yang

Wei

et al. 2023

IET Cyber-Syst and Robotics

View full text Add to dashboard Cite

show abstract

“…Loss Function. For the second stage, the overall loss for training our model can be expressed by: are the ratio-preserving losses defined in BHRL (Yang et al 2022) for evaluating the coarse novelclass and final classification results, respectively. G denotes the ground-truth label.…”

Section: Base-class Suppressionmentioning

confidence: 99%

“…In recent years, the field of few-shot object detection has thrived (Fan et al 2020a;Han et al 2022), in which prevalent approaches often incorporate transfer-learning, meta-learning, and metric-learning to deal with the task. Most existing works in OSOD adopt the metric-learning paradigm and recognize new objects based on the similarity metrics between image pairs without finetuning (Hsieh et al 2019;Zhang et al 2022b;Yang et al 2022). However, they are generally dedicated to exploring effective cross-image feature correlation to better use the limited information, neglecting the phenomenon of the model bias towards the base classes and the generalization degradation on the novel classes (Fan et al 2021;Lang et al 2022).…”

Section: Introductionmentioning

confidence: 99%

Exploring Base-Class Suppression with Prior Guidance for Bias-Free One-Shot Object Detection

Zhang,

Hu,

Shan

et al. 2024

AAAI

View full text Add to dashboard Cite

One-shot object detection (OSOD) aims to detect all object instances towards the given category specified by a query image. Most existing studies in OSOD endeavor to establish effective cross-image correlation with limited query information, however, ignoring the problems of the model bias towards the base classes and the generalization degradation on the novel classes. Observing this, we propose a novel algorithm, namely Base-class Suppression with Prior Guidance (BSPG) network to achieve bias-free OSOD. Specifically, the objects of base categories can be detected by a base-class predictor and eliminated by a base-class suppression module (BcS). Moreover, a prior guidance module (PG) is designed to calculate the correlation of high-level features in a non-parametric manner, producing a class-agnostic prior map with unbiased semantic information to guide the subsequent detection process. Equipped with the proposed two modules, we endow the model with a strong discriminative ability to distinguish the target objects from distractors belonging to the base classes. Extensive experiments show that our method outperforms the previous techniques by a large margin and achieves new state-of-the-art performance under various evaluation settings.

show abstract

“…Only with few images per new object, this approach can find unseen objects 9,10 . Indeed, the literature [11][12][13][14][15] lays the foundations of the LSOD task. Unlike conventional object detectors, this task refers to the classification and detection of objects with few information.…”

Section: Introductionmentioning

confidence: 99%

A fusion Siamese-ResNet network applied for low shot object detection in warehouses

Desmarescaux,

Kaddah,

Alfalou

et al. 2023

Pattern Recognition and Tracking XXXIV

View full text Add to dashboard Cite

Warehouses are a storage areas with high flow of goods. As part of the robotization of these areas, one of the major problems, attracting the attention of researchers, is defined by the automation of dispatching and renewing tasks when new products arrive. Indeed, having few information, automatic detection of these new products requires an update of the intelligent system, i.e. a new training step and therefore a stop in the warehouse production line. According to the literature, a novel branch of computer vision that consists in identifying and locating objects in an image with few information is called Low Shot Object Detection (LSOD). Using this solution, neural networks can automatically find new products with no need of any additional training step. To do so, neural networks architectures have been evolved by merging extracted features. This article presents a novel method that consists in merging layers convolution Siamese-ResNet network to include new products.

show abstract

Balanced and Hierarchical Relation Learning for One-shot Object Detection

Cited by 10 publications

References 18 publications

Few‐shot object detection via class encoding and multi‐target decoding

Few‐shot object detection via class encoding and multi‐target decoding

Exploring Base-Class Suppression with Prior Guidance for Bias-Free One-Shot Object Detection

A fusion Siamese-ResNet network applied for low shot object detection in warehouses

Contact Info

Product

Resources

About