The ability to identify and temporally segment finegrained human actions throughout a video is crucial for robotics, surveillance, education, and beyond. Typical approaches decouple this problem by first extracting local spatiotemporal features from video frames and then feeding them into a temporal classifier that captures high-level temporal patterns. We introduce a new class of temporal models, which we call Temporal Convolutional Networks (TCNs), that use a hierarchy of temporal convolutions to perform fine-grained action segmentation or detection. Our Encoder-Decoder TCN uses pooling and upsampling to efficiently capture long-range temporal patterns whereas our Dilated TCN uses dilated convolutions. We show that TCNs are capable of capturing action compositions, segment durations, and long-range dependencies, and are over a magnitude faster to train than competing LSTM-based Recurrent Neural Networks. We apply these models to three challenging fine-grained datasets and show large improvements over the state of the art.

The dominant paradigm for video-based action segmentation is composed of two steps: first, for each frame, compute low-level features using Dense Trajectories or a Convolutional Neural Network that encode spatiotemporal information locally, and second, input these features into a classifier that captures high-level temporal relationships, such as a Recurrent Neural Network (RNN). While often effective, this decoupling requires specifying two separate models, each with their own complexities, and prevents capturing more nuanced long-range spatiotemporal relationships. We propose a unified approach, as demonstrated by our Temporal Convolutional Network (TCN), that hierarchically captures relationships at low-, intermediate-, and high-level time-scales. Our model achieves superior or competitive performance using video or sensor data on three public action segmentation datasets and can be trained in a fraction of the time it takes to train an RNN.

The discriminative power of modern deep learning models for 3D human action recognition is growing ever so potent. In conjunction with the recent resurgence of 3D human action representation with 3D skeletons, the quality and the pace of recent progress have been significant. However, the inner workings of state-of-the-art learning based methods in 3D human action recognition still remain mostly black-box. In this work, we propose to use a new class of models known as Temporal Convolutional Neural Networks (TCN) for 3D human action recognition. Compared to popular LSTM-based Recurrent Neural Network models, given interpretable input such as 3D skeletons, TCN provides us a way to explicitly learn readily interpretable spatio-temporal representations for 3D human action recognition. We provide our strategy in re-designing the TCN with interpretability in mind and how such characteristics of the model is leveraged to construct a powerful 3D activity recognition method. Through this work, we wish to take a step towards a spatio-temporal model that is easier to understand, explain and interpret. The resulting model, Res-TCN, achieves state-of-the-art results on the largest 3D human action recognition dataset, NTU-RGBD.

Joint segmentation and classification of fine-grained actions is important for applications of human-robot interaction, video surveillance, and human skill evaluation. However, despite substantial recent progress in large-scale action classification, the performance of stateof-the-art fine-grained action recognition approaches remains low. We propose a model for action segmentation which combines low-level spatiotemporal features with a high-level segmental classifier. Our spatiotemporal CNN is comprised of a spatial component that uses convolutional filters to capture information about objects and their relationships, and a temporal component that uses large 1D convolutional filters to capture information about how object relationships change across time. These features are used in tandem with a semi-Markov model that models transitions from one action to another. We introduce an efficient constrained segmental inference algorithm for this model that is orders of magnitude faster than the current approach. We highlight the effectiveness of our Segmental Spatiotemporal CNN on cooking and surgical action datasets for which we observe substantially improved performance relative to recent baseline methods.

Interventional applications of photoacoustic imaging typically require visualization of point-like targets, such as the small, circular, cross-sectional tips of needles, catheters, or brachytherapy seeds. When these point-like targets are imaged in the presence of highly echogenic structures, the resulting photoacoustic wave creates a reflection artifact that may appear as a true signal. We propose to use deep learning techniques to identify these types of noise artifacts for removal in experimental photoacoustic data. To achieve this goal, a convolutional neural network (CNN) was first trained to locate and classify sources and artifacts in pre-beamformed data simulated with -Wave. Simulations initially contained one source and one artifact with various medium sound speeds and 2-D target locations. Based on 3,468 test images, we achieved a 100% success rate in classifying both sources and artifacts. After adding noise to assess potential performance in more realistic imaging environments, we achieved at least 98% success rates for channel signal-to-noise ratios (SNRs) of -9dB or greater, with a severe decrease in performance below -21dB channel SNR. We then explored training with multiple sources and two types of acoustic receivers and achieved similar success with detecting point sources. Networks trained with simulated data were then transferred to experimental waterbath and phantom data with 100% and 96.67% source classification accuracy, respectively (particularly when networks were tested at depths that were included during training). The corresponding mean ± one standard deviation of the point source location error was 0.40 ± 0.22 mm and 0.38 ± 0.25 mm for waterbath and phantom experimental data, respectively, which provides some indication of the resolution limits of our new CNN-based imaging system. We finally show that the CNN-based information can be displayed in a novel artifact-free image format, enabling us to effectively remove reflection artifacts from photoacoustic images, which is not possible with traditional geometry-based beamforming.