Spatio-Temporal Attention Models for Grounded Video Captioning

Mihai Zanfir, Elisabeta Marinoiu, Cristian Sminchisescu

(Submitted on 17 Oct 2016 (v1), last revised 18 Oct 2016 (this version, v2))

Automatic video captioning is challenging due to the complex interactions in dynamic real scenes. A comprehensive system would ultimately localize and track the objects, actions and interactions present in a video and generate a description that relies on temporal localization in order to ground the visual concepts. However, most existing automatic video captioning systems map from raw video data to high level textual description, bypassing localization and recognition, thus discarding potentially valuable information for content localization and generalization. In this work we present an automatic video captioning model that combines spatio-temporal attention and image classification by means of deep neural network structures based on long short-term memory. The resulting system is demonstrated to produce state-of-the-art results in the standard YouTube captioning benchmark while also offering the advantage of localizing the visual concepts (subjects, verbs, objects), with no grounding supervision, over space and time.

Comments:To appear in Asian Conference on Computer Vision (ACCV), Taipei, Taiwan, 2016Subjects:Computer Vision and Pattern Recognition (cs.CV)Cite as:arXiv:1610.04997 [cs.CV] (or arXiv:1610.04997v2 [cs.CV] for this version)

Submission history

From: Mihai Zanfir [view email] 
[v1] Mon, 17 Oct 2016 08:05:12 GMT (6260kb,D)
[v2] Tue, 18 Oct 2016 08:27:23 GMT (6260kb,D)