Detect presence of logos, text, objects, scenes, activities within video frames 

Overview  

Online analysis of a Brand based on text posts and image captions does not provide a complete social presence. Visual Insights, now with the inclusion of running on Videos, ensures comprehensive insights on multimedia posts.

​

You can now capture your Brand logos’ presence within the frames of a video. Additionally, all key visual enrichments like Objects, Scenes, Activities, Optical Character Recognition (OCR) and Inappropriate Content detection will now work on video posts.  

While in the past, logo detection was only available for X (Twitter) as a source. The new video detection model analyses key frames throughout video content across Listening sources which can help you unlock use-cases like:  

1. Matching of logos in audience, sponsor and retailer video posts across social and web data for Brand Reputation analysis   

2. Detection of specific keywords being mentioned within video content using Optical Character Recognition (OCR)   

3. Ensuring no inappropriate videos are visible to users of internal and external Sprinklr Insights reports   

4. Identifying if the Brand is being depicted in a video containing specific scenes, activities or objects   

Which visual detections are available on Videos?  

The detections in visual insights can be of 2 types. The scope for video support for all detections is detailed below –  

1. Data Fetching – These are the detections which can be added as criteria in a ‘Visual Listening’ Topic. These can be used to fetch new mentions in the partner environment.

   ​Videos are now supported for the following Data Fetching detections:

   1. ​Logo detection​

   2. ​OCR detection

2. Data Enrichments – These are the detections which are done on already fetched multimedia in Listening. These are not used to fetch new mentions in the partner environment.  

   Videos are now supported for the following Data Enrichment detections:

   1. Objects

   2. Scenes

   3. Activities

   4. Inappropriate Content 

Which sources are supported for Visual Insights on Listening?  

The support for Visual Insights on videos depends on whether the given video is supported as play-able within Sprinklr or not. Listening sources which redirect users to an external video player will not support frame-level video detections.  

A summary of support for Listening sources is given in the table below:  

Legend 

Source Details 

How to view video detections in a dashboard?  

1. Check this article to know how to setup a custom Visual Insights dashboard 

2. Choose from the available dimensions given below to plot in a required widget as per your use-case: 

*Photo Text and Photo Inappropriate Content dimensions are only available as filters and cannot be plotted in a widget