ABSTRACT
Group-activity scene graph (GASG) generation is a challenging task in computer vision, aiming to anticipate and describe relationships between subjects and objects in video sequences. Traditional video scene graph generation (VidSGG) methods focus on retrospective analysis, limiting their predictive capabilities. To enrich the scene-understanding capabilities, we introduced a GASG dataset extending the JRDB dataset with nuanced annotations involving appearance, interaction, position, relationship, and situation attributes. This work also introduces an innovative approach, a Hierarchical Attention-Flow (HAtt-Flow) mechanism, rooted in flow network theory to enhance GASG performance. Flow-attention incorporates flow conservation principles, fostering competition for sources and allocation for sinks, effectively preventing the generation of trivial attention. Our proposed approach offers a unique perspective on attention mechanisms, where conventional "values" and "keys" are transformed into sources and sinks, respectively, creating a novel framework for attention-based models. Through extensive experiments, we demonstrate the effectiveness of our Hatt-Flow model and the superiority of our proposed flow-attention mechanism. This work represents a significant advancement in predictive video scene understanding, providing valuable insights and techniques for applications that require real-time relationship prediction in video data.
ABSTRACT
BACKGROUND: As of November 15, 2013, 22 states had passed laws explicitly addressing youth access to electronic cigarettes (e-cigarettes); by 2014, this increased to 41 states. Also in 2014, more than 13.4% of youth in the U.S. reported using e-cigarettes, making e-cigarette use more prevalent than conventional cigarette use (9.2%). METHODS: We examined 221 bills addressing youth access and exposure to e-cigarettes between January 1 and November 1, 2015. Text searches on individual state general assembly websites and secondary sources were employed for data collection. Laws were analyzed using seven measures identified to protect adolescents from nicotine initiation and use. RESULTS: Two states (MI, PA) and Washington D.C. do not regulate the sale or distribution of e-cigarettes to youth as of November 1, 2015. Additionally, seventeen states have passed laws requiring e-cigarettes to use child-safety packaging to minimize unintended poisoning. As of July 1, 2016, four states (KS, LA, MN, and NC) will tax e-cigarettes. Oregon prohibits the use of e-cigarettes in cars with children under 18years of age, and Wyoming requires the public health department to develop educational campaigns to better educate the state on the risks of nicotine and tobacco products. CONCLUSION: While states are closing the gap of youth nicotine exposure, there remains a need to protect youth from e-cigarettes access, which can cause adverse health effects of brain development, lung function and potentially lead to addiction. Recommendation for the FDA to regulate e-cigarettes federally would close this regulation gap and protect youth across the U.S.
