On the Effect of Data Imbalance for Multi-Label Pedestrian Attribute Recognition

Computer Science

• Feature Extraction 100%
• Deep Learning 100%
• Recognition Performance 100%
• Convolutional Neural Network 100%
• Multiple Attribute 100%
• Evaluation Metric 100%
• Person Re-Identification 100%

Keyphrases

• Imbalanced Data 100%
• Pedestrian Attribute Recognition 100%
• Feature Extraction 33%
• Recognition Performance 33%
• Deep Learning 33%
• Adjustment Method 33%
• Attribute-based 33%
• Training Procedure 33%
• Multiple Attributes 33%
• Recent Trends 33%
• Evaluation Metrics 33%
• Person Re-identification 33%
• Two-stage Training 33%
• Balanced Data 33%
• Data Imbalance Problem 33%
• Cost Adjustment 33%