TikTok Vape Object Detection Annotations for Computer Vision Models

Object detection code and annotations developed to identify e-cigarette content in TikTok videos. The YoloV7 model was employed to detect vape devices, vapor clouds, and individuals using e-cigarettes from screenshots captured from videos with related hashtags. The code and annotation data were authored by Tanvi Anand and last updated in March 2024.

Use Cases

• Train object detection models to identify vape devices in social media images based on the described bounding box annotations.
• Analyze the prevalence and presentation of e-cigarette content on TikTok based on the described screenshot collection method.
• Benchmark model performance on challenging entities like vapor clouds and people using e-cigarettes as described.
• Develop automated content flagging systems for platforms based on the described detection of e-cigarette-specific images.

Strengths

• Annotations include bounding boxes for challenging entities like vapor clouds and individuals using e-cigarettes.
• Model development employed the YoloV7 architecture, a modern object detection framework.
• Data source is specified as TikTok videos captured via e-cigarette-related hashtags.

Limitations

• Description metadata is limited; actual data quality and file formats require manual inspection after download.
• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Texas Data Repository Harvested Dataverse

Collection Method

Screenshots captured from TikTok videos with e-cigarette-related hashtags, annotated with bounding boxes.

Time Range

null

Freshness

Last updated 2024-03-18 10:27:27; freshness should be verified.

Geography

null