2021 Tencent Advertising Algorithm Competition: Winners, Accepted Papers, and Reviewer Feedback
The 2021 Tencent Advertising Algorithm Competition, held as the ACM MM 2021 Grand Challenge, announced the top three teams for two tracks, presented the accepted multimodal video advertising papers with detailed reviewer comments, and highlighted the significance of algorithmic innovation over ranking alone.
The Tencent Advertising Algorithm Competition, organized by Tencent Advertising since 2017, partnered with the ACM Multimedia (ACM MM) 2021 Grand Challenge to focus on deep video content understanding. The 2021 edition ran from April to early July, culminating in a live finals event on August 16 where rankings for two tracks were announced.
Track 1 – Video Advertisement Millisecond Semantic Parsing winners: Champion – “GZ” team (US$100,000), Runner‑up – “swg‑lhl” team (US$20,000), Third place – “FVL” team (US$10,000). The accepted paper titled Multi‑modal Representation Learning for Video Advertisement Content Structuring from the team GZ (Sun Yat‑sen University) received reviewer praise for its multi‑stage solution, multimodal feature extractors resembling Inception, cross‑modal Transformer fusion, enhanced BMN loss, and clear, reproducible presentation.
Track 2 – Multimodal Video Advertisement Tagging winners: Champion – “Life of Extravagance” team (US$10,000), Runner‑up – “Quietly Getting Rich ⊙‑⊙” team (US$5,000), Third place – “trans_video” team (US$2,500). The accepted paper A Multimodal Framework for Video Ads Understanding from team FVL (Fudan University) was commended for strong multimodal feature extractors, OCR‑derived text, novel local‑global attention linking scene text with BERT‑based ASR, and solid experimental validation.
Additional accepted papers include A Solution to Multi‑modal Ads Video Tagging Challenge (teams Lingeo and Quietly Getting Rich) and Rethinking The Impacts Of Overfitting And Feature Quality On Small‑scale Video Classification (Pytorch Forever team). Reviewers highlighted innovative model designs, effective use of pre‑trained EfficientNet, TSM temporal modeling, and strategies to mitigate over‑fitting, while noting some typographical errors and areas needing deeper explanation.
The organizers emphasized that paper acceptance depends more on algorithmic novelty and insight than competition ranking, and invited authors to present at ACM MM 2021. The competition concluded on August 16, 2021, with a call for the 2022 edition to continue exploring cutting‑edge advertising algorithms.
Tencent Advertising Technology
Official hub of Tencent Advertising Technology, sharing the team's latest cutting-edge achievements and advertising technology applications.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.