Naver at ActivityNet Challenge 2019 -- Task B Active Speaker Detection (AVA)

Abstract

This report describes our submission to the ActivityNet Challenge at CVPR 2019. We use a 3D convolutional neural network (CNN) based front-end and an ensemble of temporal convolution and LSTM classifiers to predict whether a visible person is speaking or not. Our results show significant improvements over the baseline on the AVA-ActiveSpeaker dataset.

References in corpus (3)

• Deep Audio-Visual Speech Recognition
• LipNet: End-to-End Sentence-level Lipreading
• Perfect match: Improved cross-modal embeddings for audio-visual synchronisation

Cited by in corpus (13)

• Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection
• MovieNet: A Holistic Dataset for Movie Understanding
• MAAS: Multi-modal Assignation for Active Speaker Detection
• Active Speaker Detection as a Multi-Objective Optimization with Uncertainty-based Multimodal Fusion
• A Local-to-Global Approach to Multi-modal Movie Scene Segmentation
• FaVoA: Face-Voice Association Favours Ambiguous Speaker Detection
• Leveraging Visual Supervision for Array-based Active Speaker Detection and Localization
• Cross modal video representations for weakly supervised active speaker localization
• End-To-End Audiovisual Feature Fusion for Active Speaker Detection
• Look Who's Talking: Active Speaker Detection in the Wild
• PicTalky: Augmentative and Alternative Communication Software for Language Developmental Disabilities
• Active Speakers in Context
• Online Multi-modal Person Search in Videos