Abstract: In this work，we focus on short­term single object tracking，which is the most general type of tracking problems.Numerous significant trackers have been proposed over the past few decades.As we can see from these trackers，using a single type of feature to train the tracking model is a drawback，which usually leads to poor robustness in object tracking.On the other hand，the tracking result of a single tracker can sometimes be very unstable due to the complication of video sequences.Aiming at addressing these problems，we propose a collaborative tracking model fusing hybrid features in this paper.A generative model，which uses color­based features to compute the matching scores between target object and candidates，is combined with a discriminative model，which trains SVM with gray­scale，HOG and LBP features to classify the candidates as target or non­target objects.A spatial pyramid model is adopted in the feature calculation process of the generative model for a better description of the target and candidates.Then，the matching score and classification score are fused to rank the candidates.All the candidates are produced under the particle filter framework.The top­ranked candidate is regarded as the target object and used to update the training set for discriminative model as well.Besides，different update scheme and frequency are used for the two parts of the collaborative model.The ensemble does not impair the performance of the tracking precision but slightly outperform the single model.Finally，we compare our results with other trackers along with some analysis.The experimental results on visual tracking benchmark CVPR2013 demonstrate that our approach is effective for partial occlusion and background clutter in tracking，thus realizes desirable tracking performance under complex conditions.