TJU-DHD: Tianjin University Diverse High-Definition Dataset

This high-resolution dataset features annotated imagery categorized for object detection and pedestrian detection as presented in the IEEE TIP 2020 publication. It provides high-definition visual data specifically formatted for identifying diverse objects and human figures in complex scenes.

Use Cases

• Train pedestrian detection models using the pedestrian-specific bounding box annotations
• Develop multi-class object detection systems using the high-resolution image samples
• Validate the efficacy of high-definition feature maps in computer vision models

Strengths

• High-resolution image format for high-fidelity object localization
• Dual-category annotations for general object detection and pedestrian detection
• Methodology and data standards validated by the IEEE TIP 2020 publication