Video Coding for Machines: Large-Scale Evaluation of Deep Neural Networks Robustness to Compression Artifacts for Semantic Segmentation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Video Coding for Machines: Large-Scale Evaluation of Deep Neural Networks Robustness to Compression Artifacts for Semantic Segmentation

Résumé

In the Video Coding for Machines (VCM) context where visual content is compressed before being transmitted to a vision task algorithm, appropriate trade-off between the compression level and the vision task performance must be chosen. In this paper, a Deep Neural Networks (DNN) based semantic segmentation algorithm robustness to compression artifacts is evaluated with a total of 1486 different coding configurations. Results indicate the importance of using an appropriate image resolution to overcome the block-partitioning limitations in existing compression algorithms, allowing 58.3%, 49.8%, 33.5% and 24.3% bitrate savings at equivalent prediction accuracy for JPEG, JM, x265 and VVenC, respectively. Surprisingly, JPEG can achieve 73.41% bitrate reduction with the inclusion of compressed images at training time over VVC Test Model (VTM) with a DNN trained on pristine data, which implies that DNN generalization ability must not be overlooked.
Fichier principal
Vignette du fichier
Video Coding for Machines: Large-Scale Evaluation of Deep Neural Networks Robustness to Compression Artifacts for Semantic Segmentation.pdf (7.08 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03831514 , version 1 (27-10-2022)

Identifiants

  • HAL Id : hal-03831514 , version 1

Citer

Alban Marie, Karol Desnos, Luce Morin, Lu Zhang. Video Coding for Machines: Large-Scale Evaluation of Deep Neural Networks Robustness to Compression Artifacts for Semantic Segmentation. International Workshop on Multimedia Signal Processing (MMSP), Sep 2022, Shanghai, China. ⟨hal-03831514⟩
92 Consultations
63 Téléchargements

Partager

Gmail Facebook X LinkedIn More