Exploiting Image-trained CNN Architectures for Unconstrained Video Classification
This work addresses video classification for researchers and practitioners, offering incremental improvements over existing methods.
The paper tackled event detection in videos by exploring strategies using image-trained CNNs, achieving a mean average precision increase from 34.95% to 38.74% on the TRECVID MED'14 dataset and state-of-the-art performance on UCF-101.
We conduct an in-depth exploration of different strategies for doing event detection in videos using convolutional neural networks (CNNs) trained for image classification. We study different ways of performing spatial and temporal pooling, feature normalization, choice of CNN layers as well as choice of classifiers. Making judicious choices along these dimensions led to a very significant increase in performance over more naive approaches that have been used till now. We evaluate our approach on the challenging TRECVID MED'14 dataset with two popular CNN architectures pretrained on ImageNet. On this MED'14 dataset, our methods, based entirely on image-trained CNN features, can outperform several state-of-the-art non-CNN models. Our proposed late fusion of CNN- and motion-based features can further increase the mean average precision (mAP) on MED'14 from 34.95% to 38.74%. The fusion approach achieves the state-of-the-art classification performance on the challenging UCF-101 dataset.