VasileiosMezaris

deep learning video summarization explainable ai video event recognition unsupervised learning generative adversarial networks artificial intelligence tvsum summe attention mechanisms ca-sum image classification video event explanation sum-gan-aae attention auto-encoder sum-gan-sl deep neural networks video forensics disinformation video deepfake detection manipulation detection manipulation localization image forensics 360-degrees video dual softmax graph attention network cross-modal deep network gated-vigat vigat attention xai ad-hoc video search cross-modal retrieval saliency detection video objects lstm evaluation measures dual encoding network video analysis migration invid verification plugin unsupervised video summarization reverse video search fake news multimodal prompting image semantic descriptions clip embeddings vision language models minigpt-4 large multimodal models av scene discrepancies verification plug-in pgl-sum ai-based summarization annotation tool benchmarking dataset adversarial image generation evaluation framework visual explanations video deepfakes object removal/inpainting splicing copy-move forgeries late fusion early fusion srm bayar convolution noiseprint multi-modal fusion benchmarking news images data augmentation fine-tuning image-text matching equirectangular projection video understanding masked feature modelling masked image modelling newsimages task mediaeval t-times-v clip post-hoc explanation class-specific explanation class activation maps image classifier explanation cnn explanation trainable explanation frame selection policy gating mechanism bottom-up event recognition discoverability measures explaining video summarization inherent attention explainability attention mechanism dual-softmax trecvid video feature encoders video feature extractors free-text query object-level explanation frame-level explanation saliency map tame l-cam-img l-cam-fm l-cam learning-based xai bias deep convolutional neural nets class activation map dual softmax operation multiple space learning feature encoders cross-modal video retrieval text-based video search frame diversity frame uniqueness concentrated attention social media video consumption devices video adaptation online web service video smart cropping video aspect ratio transformation supervised learning positional encoding multi-head attention self-attention reinforcement learning video thumbnail selection improved marginal ranking loss hard-negative samples ad-hoc video retrieval ai explainability social media; cross-modal retrieval; concept textu video dataset clustering video cropping video aspect ratio retargeting multimedia and multimodal retrieval graph convolutional network ai4media invid video verification misinformation audiovisual concept detection geometric median eigenanalysis automatic structured pruning mobile multimedia video summarization evaluation f-score performance over random migration-related semantic concepts migration theories video retrieval semantic concepts deep network pruning class-separability criteria asymptotic filter pruning deep convolutional neural networks web demo content re-purposing content adaptation sum-gan demo video annotation video fragmentation video fragmentation and reverse web search news journalism sum-gan-vaae variational auto-encoder adversarial learning neglection criterion youtube-8m svhn cifar100 cifar10 subclass dnns neglected classes multimedia classification deep fakes artificiali intelligence