Long-term recurrent convolutional networks for visual recognition and description
Top Cited Papers
- 1 June 2015
- conference paper
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2625-2634
- https://doi.org/10.1109/cvpr.2015.7298878
Abstract
Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or “temporally deep”, are effective for tasks involving sequences, visual and otherwise. We develop a novel recurrent convolutional architecture suitable for large-scale visual learning which is end-to-end trainable, and demonstrate the value of these models on benchmark video recognition tasks, image description and retrieval problems, and video narration challenges. In contrast to current models which assume a fixed spatio-temporal receptive field or simple temporal averaging for sequential processing, recurrent convolutional models are “doubly deep” in that they can be compositional in spatial and temporal “layers”. Such models may have advantages when target concepts are complex and/or training data are limited. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Long-term RNN models are appealing in that they directly can map variable-length inputs (e.g., video frames) to variable length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent long-term models are directly connected to modern visual convnet models and can be jointly trained to simultaneously learn temporal dynamics and convolutional perceptual representations. Our results show such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined and/or optimized.Keywords
Other Versions
This publication has 19 references indexed in Scilit:
- CaffePublished by Association for Computing Machinery (ACM) ,2014
- Coherent Multi-sentence Video Description with Variable Level of DetailLecture Notes in Computer Science, 2014
- On the Properties of Neural Machine Translation: Encoder–Decoder ApproachesPublished by Association for Computational Linguistics (ACL) ,2014
- Translating Video Content to Natural Language DescriptionsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2013
- A Thousand Frames in Just a Few Words: Lingual Description of Videos through Latent Topics and Sparse Object StitchingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2013
- Dense Trajectories and Motion Boundary Descriptors for Action RecognitionInternational Journal of Computer Vision, 2013
- Towards textually describing complex video contents with audio-visual concept classifiersPublished by Association for Computing Machinery (ACM) ,2011
- Sequential Deep Learning for Human Action RecognitionLecture Notes in Computer Science, 2011
- Long Short-Term MemoryNeural Computation, 1997
- A Learning Algorithm for Continually Running Fully Recurrent Neural NetworksNeural Computation, 1989