tmozgach
commented
6 years ago First draft: Watching long videos is uninteresting and users face a challenge to focus on all the parts of the video with equal concentration levels. In case of instructional videos, it becomes more likely to lose track of some steps with varying concentration of the users. Therefore, there is a need to summarize the instructions in a succinct manner. Moreover, the visual content provides an additional level of verification to the users while performing the task by providing the outcome of each step. Given the fast-paced lives, users do not prefer to remember the details of such activities and therefore, feel the need to review the steps involved in performing a task before performing one.
Use of websites with video content is commonplace for the users to get an instruction manual for essential activities. However, the crowdsourced nature of instructional video content available on the internet poses several challenges in terms of variation in content detail as well as modality of information.
In this project, we perform an observation study on the users on specific types of cooking videos, We obtain the data from YouTube-8M dataset for Cooking show and Hairstyle categories and show these “in-the-wild” videos to the users. We utilize the formative evaluation of the way users watch the videos to perform a task and provide design recommendations for generation of video summaries.
In the user study we study the following things (1) A confidence level of users before and after watching the video for the given task. (2) We also investigate need of information of different modalities and the user preferences for the information modality. (3) Perceived difficulty of the task (4) Quantitative evaluation of the comments of the users on the level of detail in the videos
The goal of this study is to understand the requirements of users watching instructional videos and infer a notion of an ideal overview of the videos.
Problem statement:
Motivation
Objectives: