TransVG++: End-to-End Visual Grounding With Language Conditioned Vision Transformer.

In this work, we explore neat yet effective Transformer-based frameworks for visual grounding. The previous methods generally address the core problem of visual grounding, i.e., multi-modal fusion and reasoning, with manually-designed mechanisms. Such heuristic designs are not only complicated but also make models easily overfit specific data distributions. To avoid this, we first propose TransVG, which establishes multi-modal correspondences by Transformers and localizes referred regions by directly regressing box coordinates. We empirically show that complicated fusion modules can be replaced by a simple stack of Transformer encoder layers with higher performance. However, the core fusion Transformer in TransVG is stand-alone against uni-modal encoders, and thus should be trained from scratch on limited visual grounding data, which makes it hard to be optimized and leads to sub-optimal performance. To this end, we further introduce TransVG++ to make two-fold improvements. For one thing, we upgrade our framework to a purely Transformer-based one by leveraging Vision Transformer (ViT) for vision feature encoding. For another, we devise Language Conditioned Vision Transformer that removes external fusion modules and reuses the uni-modal ViT for vision-language fusion at the intermediate layers. We conduct extensive experiments on five prevalent datasets, and report a series of state-of-the-art records.

Potential Impact of FDA Flavor Enforcement Policy on Vaping Behavior on Twitter.

In January 2020, the FDA announced an electronic cigarette (e-cigarette) flavor enforcement policy to restrict the sale of all unauthorized cartridge-based flavored e-cigarettes except tobacco and menthol flavors, which was implemented on 6 February 2020. This study aimed to understand the potential influence of this policy on one vaping behavior change-quitting vaping-using Twitter data. Twitter posts (tweets) related to e-cigarettes were collected between June 2019 and October 2020 through a Twitter streaming API. Based on the geolocation and keywords related to quitting vaping, tweets mentioning quitting vaping from the US were filtered. The demographics (age and gender) of Twitter users who mentioned quitting vaping were further inferred using a deep learning algorithm (deepFace). The proportion of tweets and Twitter users mentioning quitting vaping were compared between before and after the announcement and implementation of the flavor policy. Compared to before the FDA flavor policy, the proportion of tweets (from 0.11% to 0.20% and 0.24%) and Twitter users (from 0.15% to 0.70% and 0.86%) mentioning quitting vaping were significantly higher after the announcement and implementation of the policy (p-value < 0.001). In addition, there was an increasing trend in the proportion of female and young adults (18-35 years old) mentioning quitting vaping on Twitter after the announcement and implementation of the policy compared to that before the policy. Our results showed that the FDA flavor enforcement policy did have a positive impact on quitting vaping on Twitter. Our study provides an initial evaluation of the potential influence of the FDA flavor enforcement policy on user vaping behavior.