Unlike in most previous partial video copy detection (PVCD) algorithms, where reference videos are scanned one by one, we treat the PVCD as a video search/retrieval problem. We propose a fast partial video copy detection framework in this paper. In this framework, all frame CNN features of the reference videos are organized in a KNN searchable database. Instead of scanning all reference videos, the query video segment does a fast KNN search in the global feature database. The returned results are used to generate a shortlist of candidate videos. A modified temporal network is then used to localize the copy segment in the candidate videos. Furthermore, We propose to use a transformer encoder to improve the CNN feature. We evaluate our algorithm on the VCDB dataset. Our benchmark F1 scores exceed state-of-the-art by a big margin. The speed of our algorithm is also improved significantly.

We present an automatic COVID1-19 diagnosis framework from lung CT images. The focus is on signal processing and classification on small datasets with efforts putting into exploring data preparation and augmentation to improve the generalization capability of the 2D CNN classification models. We propose a unique and effective data augmentation method using multiple Hounsfield Unit (HU) normalization windows. In addition, the original slice image is cropped to exclude background, and a filter is applied to filter out closed-lung images. For the classification network, we choose to use 2D Densenet and Xception with the feature pyramid network (FPN). To further improve the classification accuracy, an ensemble of multiple CNN models and HU windows is used. On the training/validation dataset, we achieve a patient classification accuracy of 93.39%.

Gun violence is a severe problem in the world, particularly in the United States. Deep learning methods have been studied to detect guns in surveillance video cameras or smart IP cameras and to send a real-time alert to security personals. One problem for the development of gun detection algorithms is the lack of large public datasets. In this work, we first publish a dataset with 51K annotated gun images for gun detection and other 51K cropped gun chip images for gun classification we collect from a few different sources. To our knowledge, this is the largest dataset for the study of gun detection. This dataset can be downloaded at www.linksprite.com/gundetection-datasets. We present a gun detection system using a smart IP camera as an embedded edge device and a cloud server as a manager for device, data, alert, and to further reduce the false positive rate. We study to find solutions for gun detection in an embedded device and for gun classification on the edge device and the cloud server. This edge/cloud framework makes the deployment of gun detection in the real world possible.

The outbreak of COVID-19 corona virus in late 2019 has put a tremendous threat to the whole world. Driven by the mission to save lives, we develop a face recognition system to detect patients with fever symptom and to trace close contacts. A real-time alert is sent to the account manager on web or mobile app to enable further actions to quarantine the patients and close contacts. This provides a way to control the spread of the virus.

Conventional method for media defect recovery in a low-density parity-check (LDPC) coded magnetic recording channel is erasure decoding. In erasure decoding, read back signal in media defect region is erased and not used in channel detection or iterative decoding. In this paper, a new method based on full-response reequalization, in which media defect corrupted signal is first reequalized to full response then partially used in iterative decoding, is proposed. Simulation results show significant performance improvement over conventional erasure decoding in both non-precoded and preceded channels.

A method to evaluate the performance of a low-density parity-check (LDPC) code on partial-response (PR) channels in terms of the noise threshold and decoding error is presented. Given a particular codeword or assuming an independent and uniformly distributed (i.u.d.) codeword for transmission, the density-evolution algorithm is used to compute the probability density function of messages passing in the decoding process, from which the decoding error is extracted. This estimated i.u.d. decoding error is used to approximate the decoding error of an ensemble of LDPC codes on arbitrary PR channels. Comparison with simulation results shows that it is a very good approximation for the simulated codes, provided their length is large enough.

Signal-to-noise ratio (SNR) mismatch has been found to have a great impact on the performance of low-density parity-check (LDPC) coded magnetic recording channels. While an inappropriate SNR mismatch degrades the performance dramatically, a properly selected optimum SNR mismatch can improve it significantly. We analyze the causes of this phenomenon and find optimum values for the SNR mismatch for specific magnetic recording systems using both density evolution analysis and Monte Carlo simulations.