filmov
tv
Human Detection, Tracking and Segmentation in Surveillance Video
Показать описание
Final Oral Examination of:
Guang Shu
For the Degree of:
Doctor of Philosophy (Computer Engineering)
This dissertation deals with the means of improving the current state of human detection, tracking and segmentation based on learning scene-specific information in a video. In this approach a DPM human detector is employed to collect detection examples; a support vector machine (SVM) classifier is trained using superpixel-based Bag-of-Words (BoW).
Given robust human detection our approach learns part-based person-specific SVM classifiers which capture articulations of moving human bodies with dynamically changing backgrounds, which handles occlusions in both detection and tracking, Next we separate superpixels corresponding to a human and background by using part-based detection models and minimizing an energy function using Conditional Random Field (CRF). Then the spatio-temporal constraints of the video is leveraged to build tracklet-based Gaussian Mixture Models (GMM), and thte boundaries are smoothed by multi-frame graph optimization.
Finally we developed an efficient real time tracking system, NONA, for high-definition surveillance video. WE implement the system using Intel Threading Building Blocks (TBB), which executes video ingestion, tracking and video output in parallel. We employ a Fast Fourier Transform based normalized cross correlation as the core tracking algorithm for efficiency. We also incorporate Adaptive Template scaling and Local Frame Difference.
Guang Shu
For the Degree of:
Doctor of Philosophy (Computer Engineering)
This dissertation deals with the means of improving the current state of human detection, tracking and segmentation based on learning scene-specific information in a video. In this approach a DPM human detector is employed to collect detection examples; a support vector machine (SVM) classifier is trained using superpixel-based Bag-of-Words (BoW).
Given robust human detection our approach learns part-based person-specific SVM classifiers which capture articulations of moving human bodies with dynamically changing backgrounds, which handles occlusions in both detection and tracking, Next we separate superpixels corresponding to a human and background by using part-based detection models and minimizing an energy function using Conditional Random Field (CRF). Then the spatio-temporal constraints of the video is leveraged to build tracklet-based Gaussian Mixture Models (GMM), and thte boundaries are smoothed by multi-frame graph optimization.
Finally we developed an efficient real time tracking system, NONA, for high-definition surveillance video. WE implement the system using Intel Threading Building Blocks (TBB), which executes video ingestion, tracking and video output in parallel. We employ a Fast Fourier Transform based normalized cross correlation as the core tracking algorithm for efficiency. We also incorporate Adaptive Template scaling and Local Frame Difference.
Комментарии