[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-31874":3,"doc-seo-31874":27},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"file_id":15,"file_url":16,"file_type":17,"file_size":18,"view_count":4,"is_deleted":4,"is_public":19,"is_downloadable":19,"audit_status":19,"page_count":20,"language":21,"language_code":22,"table_of_contents":23,"faqs":24,"seo_title":13,"seo_description":14,"update_tm":25,"read_time":26},31874,13056703019404,"Miles","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","LW-DeepFakeNet: a Lightweight Time Distributed CNN-LSTM Network for Real-Time DeepFake Video Detection","Large-scale social media and generative deep learning have enabled highly realistic misleading videos known as DeepFakes, creating serious challenges for authenticity verification. LW-DeepFakeNet is proposed as a lightweight time distributed CNN-LSTM model that jointly uses spatial and temporal cues to classify altered videos. Transfer learning leverages pre-trained convolutional networks for spatial feature extraction, while LSTMs capture temporal dynamics with limited training data and time. A scenario involving scene changes is handled via class-imbalance countermeasures, achieving up to 152× fewer parameters, 99.24% accuracy, and about 80 fps.","cbCaijTBBCKGwNr4","https://ap.wps.com/l/cbCaijTBBCKGwNr4","pdf",1121575,1,9,"English","en","# Introduction\n## DeepFake video verification challenges\n## Spatial vs. spatio-temporal detection approaches\n## Proposed LW-DeepFakeNet approach","[{\"question\":\"What problem does LW-DeepFakeNet address?\",\"answer\":\"It targets real-time detection of DeepFake videos by determining whether a video has been altered, even when edited content includes noise or changing scenes.\"},{\"question\":\"How does LW-DeepFakeNet combine spatial and temporal information?\",\"answer\":\"It uses a pre-trained VGG16 backbone for spatial feature extraction and an LSTM layer to learn sequential temporal features across frames.\"},{\"question\":\"How does the model improve efficiency and parameter usage?\",\"answer\":\"The spatial feature extractor is frozen, so training focuses mainly on the temporal component, leading to a lightweight model with up to 152× fewer parameters and about 80 fps performance.\"}]",1780347805,23,{"code":4,"msg":28,"data":29},"ok",{"site_id":30,"language":22,"slug":31,"title":13,"keywords":32,"description":14,"schema_data":33,"social_meta":84,"head_meta":86,"extra_data":88,"updated_unix":25},105,"lw-deepfakenet-a-lightweight-time-distributed-cnn-lstm-network-for-real-time-deepfake-video-detection","",{"@graph":34,"@context":83},[35,52,66],{"@type":36,"itemListElement":37},"BreadcrumbList",[38,42,46,49],{"item":39,"name":40,"@type":41,"position":19},"https://docshare.wps.com","Home","ListItem",{"item":43,"name":44,"@type":41,"position":45},"https://docshare.wps.com/document/","Document",2,{"item":47,"name":12,"@type":41,"position":48},"https://docshare.wps.com/document/research-report/",3,{"item":50,"name":13,"@type":41,"position":51},"https://docshare.wps.com/document/lw-deepfakenet-a-lightweight-time-distributed-cnn-lstm-network-for-real-time-deepfake-video-detection/31874/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"description":14,"dateModified":60,"datePublished":60,"encodingFormat":59,"isAccessibleForFree":61,"interactionStatistic":62},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":39,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-06-01",true,{"@type":63,"interactionType":64,"userInteractionCount":4},"InteractionCounter",{"@type":65},"ViewAction",{"@type":67,"mainEntity":68},"FAQPage",[69,75,79],{"name":70,"@type":71,"acceptedAnswer":72},"What problem does LW-DeepFakeNet address?","Question",{"text":73,"@type":74},"It targets real-time detection of DeepFake videos by determining whether a video has been altered, even when edited content includes noise or changing scenes.","Answer",{"name":76,"@type":71,"acceptedAnswer":77},"How does LW-DeepFakeNet combine spatial and temporal information?",{"text":78,"@type":74},"It uses a pre-trained VGG16 backbone for spatial feature extraction and an LSTM layer to learn sequential temporal features across frames.",{"name":80,"@type":71,"acceptedAnswer":81},"How does the model improve efficiency and parameter usage?",{"text":82,"@type":74},"The spatial feature extractor is frozen, so training focuses mainly on the temporal component, leading to a lightweight model with up to 152× fewer parameters and about 80 fps performance.","https://schema.org",{"og:url":50,"og:type":85,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":87,"canonical":50},"index,follow",{"doc_id":7,"site_id":30}]