FSSD: Feature Fusion Single Shot Multibox Detector

Li, Zuoxin; Zhou, Fuqiang

doi:10.48550/arxiv.1712.00960

Cited by 178 publications

(175 citation statements)

References 28 publications

Supporting

Mentioning

175

Contrasting

Order By: Relevance

“…To get an MTL network that is lightweight In the feature extraction part, the Res 1 to Res 4 are generated by ResNet50 and Res 5 is an additional layer. Then the feature fusion technology similar to FSSD [21] is applied to fuse the Res 2 to Res 5 into one feature F 0 . The dimension of F 0 is (64, 64, 512) and is represented by (height, width, channels).…”

Section: B Feature Fusion Based Multi-task Network (Ffmnet)mentioning

confidence: 99%

Deep Joint Source-Channel Coding for Multi-Task Network

Wang,

Zhang,

et al. 2021

Preprint

View full text Add to dashboard Cite

Multi-task learning (MTL) is an efficient way to improve the performance of related tasks by sharing knowledge. However, most existing MTL networks run on a single end and are not suitable for collaborative intelligence (CI) scenarios. In this work, we propose an MTL network with a deep joint sourcechannel coding (JSCC) framework, which allows operating under CI scenarios. We first propose a feature fusion based MTL network (FFMNet) for joint object detection and semantic segmentation. Compared with other MTL networks, FFMNet gets higher performance with fewer parameters. Then FFMNet is split into two parts, which run on a mobile device and an edge server respectively. The feature generated by the mobile device is transmitted through the wireless channel to the edge server. To reduce the transmission overhead of the intermediate feature, a deep JSCC network is designed. By combining two networks together, the whole model achieves 512× compression for the intermediate feature and a performance loss within 2% on both tasks. At last, by training with noise, the FFMNet with JSCC is robust to various channel conditions and outperforms the separate source and channel coding scheme.

show abstract

Section: B Feature Fusion Based Multi-task Network (Ffmnet)mentioning

confidence: 99%

Deep Joint Source-Channel Coding for Multi-Task Network

Wang,

Zhang,

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…On the other hand, YOLO series [20][21][22] fall into the second branch, which tackles object detection as an end-to-end regression problem. Besides, the known SSD series [15,19] propose to utilize pre-defined bounding boxes to adjust to various object scales inspired by [23].…”

Section: General Object Detectionmentioning

confidence: 99%

“…Multi-scale features have been exhaustively exploited for multi-scale objects in general object detection [14-16, 19, 21, 25]. For example, FSSD [15] proposed to fuse multiscale feature and implement detection on the fused feature map. Lin et al constructed the feature pyramid network (FPN) [16], which builds a top-down architecture and employs multi-scale feature map for detection.…”

Section: Multi-scale Feature Extractionmentioning

confidence: 99%

AirDet: Few-Shot Detection without Fine-tuning for Autonomous Exploration

Li¹,

Wang²,

Reddy³

et al. 2021

Preprint

View full text Add to dashboard Cite

Few-shot object detection has rapidly progressed owing to the success of meta-learning strategies. However, the requirement of a fine-tuning stage in existing methods is timeconsuming and significantly hinders their usage in real-time applications such as autonomous exploration of low-power robots. To solve this problem, we present a brand new architecture, AirDet, which is free of fine-tuning by learning class agnostic relation with support images. Specifically, we propose a support-guided cross-scale (SCS) feature fusion network to generate object proposals, a global-local relation network (GLR) for shots aggregation, and a relation-based prototype embedding network (R-PEN) for precise localization. Exhaustive experiments are conducted on COCO and PASCAL VOC datasets, where surprisingly, AirDet achieves comparable or even better results than the exhaustively finetuned methods, reaching up to 40-60% improvements on the baseline. To our excitement, AirDet obtains favorable performance on multi-scale objects, especially the small ones. Furthermore, we present evaluation results on real-world exploration tests from the DARPA Subterranean Challenge, which strongly validate the feasibility of AirDet in robotics. The source code, pre-trained models, along with the real world data for exploration, will be made public.

show abstract

“…Kisantal et al [14] proposed a data augmentation method by oversample images with small objects and copy-pasting small objects many times. Li et al [15] presented feature fusion single shot multibox detector (FSSD), which adopted a new lightweight feature fusion module which improved the small object detection performance over SSD. Deng et al [16] proposed the extended feature pyramid network (EFPN), which introduced an extra high-resolution pyramid level specialized for small object detection.…”

Section: Introductionmentioning

confidence: 99%

A Comprehensive Approach for UAV Small Object Detection with Simulation-based Transfer Learning and Adaptive Fusion

Rui,

Youwei,

Huafei

et al. 2021

Preprint

View full text Add to dashboard Cite

Precisely detection of Unmanned Aerial Vehicles(UAVs) plays a critical role in UAV defense systems. Deep learning is widely adopted for UAV object detection whereas researches on this topic are limited by the amount of dataset and small scale of UAV. To tackle these problems, a novel comprehensive approach that combines transfer learning based on simulation data and adaptive fusion is proposed. Firstly, the open-source plugin AirSim proposed by Microsoft is used to generate mass realistic simulation data. Secondly, transfer learning is applied to obtain a pre-trained YOLOv5 model on the simulated dataset and fine-tuned model on the real-world dataset. Finally, an adaptive fusion mechanism is proposed to further improve small object detection performance. Experiment results demonstrate the effectiveness of simulation-based transfer learning which leads to a 2.7% performance increase on UAV object detection. Furthermore, with transfer learning and adaptive fusion mechanism, 7.1% improvement is achieved compared to the original YOLO v5 model.

show abstract

FSSD: Feature Fusion Single Shot Multibox Detector

Cited by 178 publications

References 28 publications

Deep Joint Source-Channel Coding for Multi-Task Network

Deep Joint Source-Channel Coding for Multi-Task Network

AirDet: Few-Shot Detection without Fine-tuning for Autonomous Exploration

A Comprehensive Approach for UAV Small Object Detection with Simulation-based Transfer Learning and Adaptive Fusion

Contact Info

Product

Resources

About