Background: Ultra-processed food consumption is a risk factor for obesity and has a negative environmental impact. Food companies spend billions of dollars on advertisements each year to increase the consumption of ultra-processed food. In Australia, USA, and New Zealand, most food advertisements around schools and in train stations promote ultra-processed food, but no similar studies have been conducted in Sweden. The aim of this study was to explore the proportion of ultra-processed food advertisements in two districts of Stockholm, Sweden with low vs. high socioeconomic status (SES). Methods: Two independent researchers (per area) mapped all advertisements, including storefronts, in two Stockholm districts. During consecutive days, all advertisements were photographed in Skärholmen (low SES district), and Östermalmstorg (high SES district), on the streets inside and outside the subway stations, as well as inside and outside of local shopping malls. Advertisements promoting food products were identified and a trained dietician categorized whether they promoted ultra-processed foods. Chi-Square test was conducted to test for differences in the proportion of ultra-processed food advertisements between the two study areas. Results: In total, 4092 advertisements were photographed in Skärholmen (n = 1935) and Östermalm (n = 2157). 32.8% of all advertisements promoted food, while 65.4% of food advertisements promoted ultra-processed foods. A significantly higher proportion of ultra-processed food advertisements out of total food advertisements was identified in the low SES area, irrespective of the researcher taking the pictures (74.6% vs. 61.8%, p < 0.001 and 70.4% vs. 54.8%, p = 0.001). There was no significant difference in the proportion of food advertisements out of total advertisements between the two areas. Conclusions: This study provides initial evidence about the scale and the differences in exposure to food advertisements across areas in Stockholm. The observed high proportion of ultra-processed food advertisements is concerning and is in sharp contrast to the Swedish dietary guidelines that recommend reduced consumption of such foods. Based on our results, residents in low SES areas might be more exposed to ultra-processed food advertisements than those in high SES areas in Stockholm. If such findings are confirmed in additional areas, they should be considered during the deployment of food advertisement regulatory actions.

Eating behavior can have an important effect on, and be correlated with, obesity and eating disorders. Eating behavior is usually estimated through self-reporting measures, despite their limitations in reliability, based on ease of collection and analysis. A better and widely used alternative is the objective analysis of eating during meals based on human annotations of in-meal behavioral events (e.g., bites). However, this methodology is time-consuming and often affected by human error, limiting its scalability and cost-effectiveness for large-scale research. To remedy the latter, a novel “Rapid Automatic Bite Detection” (RABiD) algorithm that extracts and processes skeletal features from videos was trained in a video meal dataset (59 individuals; 85 meals; three different foods) to automatically measure meal duration and bites. In these settings, RABiD achieved near perfect agreement between algorithmic and human annotations (Cohen’s kappa κ = 0.894; F1-score: 0.948). Moreover, RABiD was used to analyze an independent eating behavior experiment (18 female participants; 45 meals; three different foods) and results showed excellent correlation between algorithmic and human annotations. The analyses revealed that, despite the changes in food (hash vs. meatballs), the total meal duration remained the same, while the number of bites were significantly reduced. Finally, a descriptive meal-progress analysis revealed that different types of food affect bite frequency, although overall bite patterns remain similar (the outcomes were the same for RABiD and manual). Subjects took bites more frequently at the beginning and the end of meals but were slower in-between. On a methodological level, RABiD offers a valid, fully automatic alternative to human meal-video annotations for the experimental analysis of human eating behavior, at a fraction of the cost and the required time, without any loss of information and data fidelity.