Intro
Preface
Organization
Contents
Computer Vision
U-YOLO: Improved YOLOv5 for Small Object Detection on UAV-Captured Images
1 Introduction
2 Related Work
2.1 Small Object Detection
2.2 Visual Attention Mechanism
3 U-YOLO
3.1 Multi-scale Feature Fusion Network Extension
3.2 Detection Head with CBAM
3.3 Context-Based Attention Feature Fusion Module
4 Experiments
4.1 Implementation Details
4.2 State-of-the-Art Comparison
4.3 Ablation Studies
5 Conclusion
References

Fruit Detection Based on Automatic Occlusion Prediction and Improved YOLOv5s
1 Introduction
2 Target Detection Based on Occlusion Information Automatic Judgment
2.1 Analysis of Occlusion Information for Target Detection
2.2 Automatically Generate Occlusion Information
3 Model and Improvement
3.1 Embedded Attention Module
3.2 Improvement of Loss Function
3.3 False Detection Handle Based on Category Unification
4 Experiment
4.1 Test Environment
4.2 Evaluation Indicators
4.3 Results and Analysis
5 Conclusion
References

A Novel Autoencoder for Task-Driven Object Segmentation
1 Introduction
2 Related Work
2.1 Salient Object Segmentation
2.2 Attention Mechanism
3 Architecture
3.1 Network Architecture Overview
3.2 Encoder Module
3.3 Decoder Module
3.4 Training
4 Experiments
4.1 Datasets
4.2 Implementation Details
4.3 Evaluation Metrics
4.4 Performance Comparison on Saliency Datasets
5 Conclusions and Future Works
References
Feedback Attention-Augmented Bilateral Network for Amodal Instance Segmentation
1 Introduction
2 Related Work

2.1 Amodal 3D Object Detection
2.2 Amodal Instance Segmentation
2.3 Attention Mechanism
2.4 Feature Fusion in Deep Learning
3 Proposed Method
3.1 Overview
3.2 Feedback Attention-Augmented Network
3.3 Spatial Detail Preservation Network
3.4 Feature Fusion Module
4 Experiments
4.1 Implementation Details
4.2 Datasets and Evaluation Metrics
4.3 Comparing with Other Methods on COCOA Dataset
4.4 Ablation Study on COCOA Dataset
4.5 Visualization Results on COCOA Dataset
4.6 Results on D2SA Dataset
5 Conclusion
References

Squeeze-and-Excitation Block Based Mask R-CNN for Object Instance Segmentation
1 Introduction
2 Related Research
3 Proposed Methodology
3.1 Outline
3.2 Squeeze-and-Excitation Block
3.3 Differences from Conventional Methods
4 Experiments
4.1 Experimental Methods
4.2 Datasets
4.3 Experimental Environment
4.4 Experimental Results
4.5 Consideration
5 Conclusion
References
PointNetX: Part Segmentation Based on PointNet Promotion
1 Introduction
2 Related Work
2.1 Projection-Biassed Approach
2.2 Networks that Deal Directly with Point Clouds