Household Pet Dangerous Behavior Recognition Dataset

With the increase in the number of household pets, safety hazards caused by pets have become increasingly serious, such as pets knocking over babies or biting electrical wires, posing a threat to family safety. Existing monitoring systems lack intelligent recognition of pet behaviors, leading to the inability to provide timely warnings and interventions. This dataset aims to improve the accuracy and real-time recognition of pet behaviors through high-quality video data, helping families to detect potential dangerous behaviors in a timely manner, thereby effectively protecting family safety. Data collection is done using actual home surveillance cameras, providing a realistic and diverse environment covering various pets and household scenarios. In terms of quality control, measures such as multiple rounds of labeling and expert review ensure consistency and accuracy of annotations. The data is stored in MP4 format, with a clear organizational structure, facilitating subsequent processing and analysis.