International Conference on Computer and Knowledge Engineering

Home / 11th International Conference on Computer and Knowledge Engineering

To Transfer or Not To Transfer (TNT): Action Recognition in Still Image Using Transfer Learning

Authors :

Ali Soltani Nezhad¹ Hojat Asgarian Dehkordi² Seyed Sajad Ashrafi³ Shahriar Baradaran Shokouhi⁴

1- School of Electrical Engineering, Iran University of Science and Technology, Tehran 16846-13144, Iran 2- School of Electrical Engineering, Iran University of Science and Technology, Tehran 16846-13144, Iran 3- School of Electrical Engineering, Iran University of Science and Technology, Tehran 16846-13144, Iran 4- School of Electrical Engineering, Iran University of Science and Technology, Tehran 16846-13144, Iran

Keywords :

Action Recognition, Still Image, Transfer Learning, Knowledge Distillation

Abstract :

Lack of training data and small datasets is one of the main challenges in human action recognition in still images. For this reason, training a network from scratch does not lead to good results just by using this little training data. Hence, many existing methods use the transfer learning techniques such as fine-tuning a pre-trained network with initial weights from ImageNet. It is important to note that some of the weights obtained from ImageNet are not suitable for human action recognition tasks. On the other hand, in fine-tuning the network, suitable initial weights for human action recognition may be changed. This paper proposes a method called To Transfer or Not To Transfer (TNT) based on knowledge distillation. In this method, a none trainable teacher with ImageNet weights is employed to train a light student network for action recognition tasks. In order to transfer relevant knowledge and not to transfer insufficient knowledge from teacher to student, a To Transfer or Not To Transfer Loss (TNTL) function is introduced in this paper. The proposed method is evaluated on Stanford 40 and Pascal VOC datasets, and the results show the superiority of this method over existing methods that exploit more parameters.