Audience measurement is one of the key challenges for several media owners to understand, in an anonymous way, the level of engagement during the user experience. Consumers today have more media and channel choices than ever before. Future SmartTV, Retail and Digital Signage sectors are just few examples where video analytics may help developing the respective markets due to nowadays lack of metrics. In fact, grabbing anonymous information from the human being is important to measure the effectiveness of a campaign/television show and to take prompt actions to maximize the attention of people to the ad/show or to the product space. Point of sale and on-shelf solutions are also getting more pervasive due to the needs of measuring how shoppers engage, where attention and gaze estimation in free environment is difficult to perform. Also, measuring the customer experience will allow to stimulate a multidisciplinary approach which will bring ethologist, psycologist, marketing and media planner professionals to eventually propose new metrics, to study and understand social behaviours of social media.
Video analytics may help understanding the effectiveness of the branded message/television show by studying and measuring public opinion and polling, geographical concentration of conversation of viewers. To this aim, computer vision and pattern recognition technologies will play an important improvement in audience measurement for its capability to understand several visual cues such as demographics, free gaze estimation, dwell time, emotion and group people proxemics, where low spatial resolution of acquired subjects, changing of the pose, occlusion, illumination changes, large variability of intra-class female age and ethnicity cohorts represent some critical aspects for recognition.
The aim of this workshop is to provide an overview of recent development, state of the art methods and systems for audience measurements in retail and Digital Signage, to attract end-users, and to stimulate the creation of appropriate benchmark dataset to be used as a reference tool for the development of novel audience measurement algorithms. Researchers will be invited to submit papers under the following topics (but not limited to):
- Dwell time estimation
- Gender recognition
- Age and Age group estimation
- Behaviour analysis
- People counting in multi-camera network
- People recurrence in long time window
- Ethnicity recognition
- Emotion analysis
- Free eye gaze estimation
- Group of related people: detection, tracking and behaviour analysis
- Path optimization and queue management in the point of sale
- Privacy preserving in audience measurement
- Modelling consumer behaviour
- Annotated Dataset proposal
- BigData analytics for audience measurement
- Multimodal approaches for audience measurement
- Other visual cues for customer profiling