Pedestrian attribute classification with multi-scale and multi-label convolutional neural networks

Pedestrian attribute classification from a pedestrian image captured in surveillance scenarios is challenging due to diverse clothing appearances,varied poses and different camera views. A multiscale and multi-label convolutional neural network( MSMLCNN) is proposed to predict multiple pedestrian attributes simultaneously. The pedestrian attribute classification problem is firstly transformed into a multi-label problem including multiple binary attributes needed to be classified. Then,the multi-label problem is solved by fully connecting all binary attributes to multi-scale features with logistic regression functions. Moreover,the multi-scale features are obtained by concatenating those featured maps produced from multiple pooling layers of the MSMLCNN at different scales. Extensive experiment results show that the proposed MSMLCNN outperforms state-of-the-art pedestrian attribute classification methods with a large margin.

Network video quality assessment based on fuzzy inference system

The objective assessment method of network video quality is a challenge, because the video quality will be distorted by various factors, including transmission and compression. In order to improve the objective method, an objective assessment method based on fuzzy inference system of Mamdani is proposed. Firstly, six quality parameters are introduced. All the quality parameters are inputted to fuzzy logic controller system. Secondly, the outputs are used as next inputs and inferred by another fuzzy logic controller system to obtain the objective quality of network video. Lastly, the performance of proposed method is validated on four videos with different network environment. Meanwhile this method is compared with other methods. The experimental results show that the proposed method can improve the similarity between subjective and objective assessment.

Quality of experience models for network video quality

Nowadays,the service of network video is increasing explosively.But the quality of experience(QoE)model of network video quality is not stable.The video quality may be impaired by many factors.This paper proposes QoE models for network video quality.It consists of two components:1)the perceptual video quality model considering the impair factors related to video content as well as distortion caused by content and transmission.Next the model is built through a decision tree using a set of measured features form the network video.This proposed model can qualitatively give the grade of video quality and improve the accuracy of prediction.2)Based on the above model,another model is proposed to give the concrete objective score of video quality.It also considers original impair factors and predicts the video quality using fuzzy decision tree.The two models have their own advantages.The first model has a good computational complexity;the second model is more precise.All the models are simulated by actual experiments.They can improve the accuracy of objective model.The detail results are shown.