MSO‐DETR: Metric space optimization for few‐shot object detection
Abstract In the metric‐based meta‐learning detection model, the distribution of training samples in the metric space has great influence on the detection performance, and this influence is usually ignored by traditional meta‐detectors. In addition, the design of metric space might be interfered with...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-12-01
|
| Series: | CAAI Transactions on Intelligence Technology |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/cit2.12342 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841543328050446336 |
|---|---|
| author | Haifeng Sima Manyang Wang Lanlan Liu Yudong Zhang Junding Sun |
| author_facet | Haifeng Sima Manyang Wang Lanlan Liu Yudong Zhang Junding Sun |
| author_sort | Haifeng Sima |
| collection | DOAJ |
| description | Abstract In the metric‐based meta‐learning detection model, the distribution of training samples in the metric space has great influence on the detection performance, and this influence is usually ignored by traditional meta‐detectors. In addition, the design of metric space might be interfered with by the background noise of training samples. To tackle these issues, we propose a metric space optimisation method based on hyperbolic geometry attention and class‐agnostic activation maps. First, the geometric properties of hyperbolic spaces to establish a structured metric space are used. A variety of feature samples of different classes are embedded into the hyperbolic space with extremely low distortion. This metric space is more suitable for representing tree‐like structures between categories for image scene analysis. Meanwhile, a novel similarity measure function based on Poincaré distance is proposed to evaluate the distance of various types of objects in the feature space. In addition, the class‐agnostic activation maps (CCAMs) are employed to re‐calibrate the weight of foreground feature information and suppress background information. Finally, the decoder processes the high‐level feature information as the decoding of the query object and detects objects by predicting their locations and corresponding task encodings. Experimental evaluation is conducted on Pascal VOC and MS COCO datasets. The experiment results show that the effectiveness of the authors’ method surpasses the performance baseline of the excellent few‐shot detection models. |
| format | Article |
| id | doaj-art-a243aab6834a4ba4b2c689acd7104f4d |
| institution | Kabale University |
| issn | 2468-2322 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Wiley |
| record_format | Article |
| series | CAAI Transactions on Intelligence Technology |
| spelling | doaj-art-a243aab6834a4ba4b2c689acd7104f4d2025-01-13T14:05:51ZengWileyCAAI Transactions on Intelligence Technology2468-23222024-12-01961515153310.1049/cit2.12342MSO‐DETR: Metric space optimization for few‐shot object detectionHaifeng Sima0Manyang Wang1Lanlan Liu2Yudong Zhang3Junding Sun4School of Computer Science and Technology Henan Polytechnic University Jiaozuo ChinaSchool of Computer Science and Technology Henan Polytechnic University Jiaozuo ChinaFaculty of Arts and Law Henan Polytechnic University Jiaozuo ChinaSchool of Computer Science and Technology Henan Polytechnic University Jiaozuo ChinaSchool of Computer Science and Technology Henan Polytechnic University Jiaozuo ChinaAbstract In the metric‐based meta‐learning detection model, the distribution of training samples in the metric space has great influence on the detection performance, and this influence is usually ignored by traditional meta‐detectors. In addition, the design of metric space might be interfered with by the background noise of training samples. To tackle these issues, we propose a metric space optimisation method based on hyperbolic geometry attention and class‐agnostic activation maps. First, the geometric properties of hyperbolic spaces to establish a structured metric space are used. A variety of feature samples of different classes are embedded into the hyperbolic space with extremely low distortion. This metric space is more suitable for representing tree‐like structures between categories for image scene analysis. Meanwhile, a novel similarity measure function based on Poincaré distance is proposed to evaluate the distance of various types of objects in the feature space. In addition, the class‐agnostic activation maps (CCAMs) are employed to re‐calibrate the weight of foreground feature information and suppress background information. Finally, the decoder processes the high‐level feature information as the decoding of the query object and detects objects by predicting their locations and corresponding task encodings. Experimental evaluation is conducted on Pascal VOC and MS COCO datasets. The experiment results show that the effectiveness of the authors’ method surpasses the performance baseline of the excellent few‐shot detection models.https://doi.org/10.1049/cit2.12342few‐shot object detectionhyperbolic spacemeta‐learningmetric space |
| spellingShingle | Haifeng Sima Manyang Wang Lanlan Liu Yudong Zhang Junding Sun MSO‐DETR: Metric space optimization for few‐shot object detection CAAI Transactions on Intelligence Technology few‐shot object detection hyperbolic space meta‐learning metric space |
| title | MSO‐DETR: Metric space optimization for few‐shot object detection |
| title_full | MSO‐DETR: Metric space optimization for few‐shot object detection |
| title_fullStr | MSO‐DETR: Metric space optimization for few‐shot object detection |
| title_full_unstemmed | MSO‐DETR: Metric space optimization for few‐shot object detection |
| title_short | MSO‐DETR: Metric space optimization for few‐shot object detection |
| title_sort | mso detr metric space optimization for few shot object detection |
| topic | few‐shot object detection hyperbolic space meta‐learning metric space |
| url | https://doi.org/10.1049/cit2.12342 |
| work_keys_str_mv | AT haifengsima msodetrmetricspaceoptimizationforfewshotobjectdetection AT manyangwang msodetrmetricspaceoptimizationforfewshotobjectdetection AT lanlanliu msodetrmetricspaceoptimizationforfewshotobjectdetection AT yudongzhang msodetrmetricspaceoptimizationforfewshotobjectdetection AT jundingsun msodetrmetricspaceoptimizationforfewshotobjectdetection |