Result: Design and Development of a Video Segmentation Service: Techniques, Integration, and Optimization for Production Use ; Design e Desenvolvimento de um Serviço de Segmentação de Video: Técnicas, Integração e Optimização para Uso em Produção

Title:
Design and Development of a Video Segmentation Service: Techniques, Integration, and Optimization for Production Use ; Design e Desenvolvimento de um Serviço de Segmentação de Video: Técnicas, Integração e Optimização para Uso em Produção
Contributors:
Gonçalves, André Perdigão da Costa de Sá, Macedo, Luís Miguel Machado Lopes
Publication Year:
2025
Collection:
Universidade de Coimbra: Estudo Geral
Document Type:
Dissertation/ Thesis master thesis
Language:
English
Rights:
info:eu-repo/semantics/openAccess
Accession Number:
edsbas.7216E656
Database:
BASE

Further Information

Dissertação de Mestrado em Engenharia Informática apresentada à Faculdade de Ciências e Tecnologia ; Social networks have become the primary platform for communication and infor-mation sharing, transforming occasional posting into full-time content creationfor millions. For brands, this shift has introduced new challenges: Instead of re-lying only on designers or external agencies, companies now depend on a widenetwork of contributors, including employees, affiliates or advocates, to increasethe scale of produced content. While this decentralized approach boosts volumeand authenticity, it also introduces two main challenges: How to provide theright tools to produce content to people with no professional background in thearea and how to ensure that the content produced aligns with brand identity andstandards?In response, sqill offers an intelligent workflow for content creation and manage-ment. Powered by artificial intelligence (AI), the mobile video editor assists usersby analyzing and evaluating visual content to ensure it meets a brand predefinedguidelines. By integrating an AI driven image analysis system, sqill helps main-tain brand consistency while empowering more people to contribute with highquality content across multiple platforms.To extend this capability beyond static images and into videos, new challengesarise. Since the AI system can only evaluate images, video content must first beunderstood and translated into representative frames that retain most contextualinformation needed for analysis.This report explores the key technologies that enable this transformation. It de-tails the development of a Python based microservice designed to segment videosinto representative frames, ensuring that each frame captures sufficient contextfor accurate and fast evaluation. Additionally, it discusses the integration of thismicroservice into sqill’s existing architecture, demonstrating how this enhance-ment enables a more robust and scalable content analysis AI supported infras-tructure. ; As redes sociais ...