We present an alternative layer to convolution layers in convolutional neural networks (CNNs). Our approach reduces the complexity of convolutions by replacing it with binary decisions. Those binary decisions are used as indexes to conditional probability distributions where each probability represents a leaf in a decision tree. This means that only the indices to the probabilities need to be determined once, thus reducing the complexity of convolutions by the depth of the output tensor. Index computation is performed by simple binary decisions that require fewer CPU cycles compared to conventionally used multiplications. In addition, we show how convolutions can be replaced by binary decisions. These binary decisions form indices in the conditional probability distributions and we show how they are used to replace 2D weight matrices as well as 3D weight tensors. These new layers can be trained like convolution layers in CNNs based on the backpropagation algorithm, for which we provide a formalization. Our results on multiple publicly available data sets show that our approach outperforms conventional CNNs. Beyond the formalized reduction of complexity and the improved qualitative performance, we show empirically a significant runtime improvement compared to convolution layers. DOWNLOAD EXAMPLES: https://drive.google.com/open?id=1gqLD5N--tqVNCixenXiptcaR5hAP5IYJ
We present a new loss function for the validation of image landmarks detected via Convolutional Neural Networks (CNNs). The network learns to estimate how accurate its landmark estimation is. This loss function is applicable to all regression-based location estimations and allows exclusion of unreliable landmarks from further processing. In addition, we formulate a novel batch balancing approach which weights the importance of samples based on their produced loss. This is done by computing a probability distribution mapping on an interval from which samples can be selected using a uniform random selection scheme. We conducted several experiments on the 300W facial landmark data. In the first experiment, the influence of our batch balancing approach is evaluated by comparing it against uniform sampling. Afterwards, we compare two networks with the state of the art and demonstrate the usage and practical importance of our landmark validation signal. The effectiveness of our validation signal is further confirmed by a correlation analysis over all landmarks. Finally, we show a study on head pose estimation of truck drivers on German highways and compare our network to a commercial multi-camera system.
Eye movements hold information about human perception, intention and cognitive state. Various algorithms have been proposed to identify and distinguish eye movements, particularly fixations, saccades, and smooth pursuits. A major drawback of existing algorithms is that they rely on accurate and constant sampling rates, impeding straightforward adaptation to new movements such as micro saccades. We propose a novel eye movement simulator that i) probabilistically simulates saccade movements as gamma distributions considering different peak velocities and ii) models smooth pursuit onsets with the sigmoid function. This simulator is combined with a machine learning approach to create detectors for general and specific velocity profiles. Additionally, our approach is capable of using any sampling rate, even with fluctuations. The machine learning approach consists of different binary patterns combined using conditional distributions. The simulation is evaluated against publicly available real data using a squared error, and the detectors are evaluated against state-of-the-art algorithms.
Real-time, accurate, and robust pupil detection is an essential prerequisite for pervasive video-based eye-tracking. However, automated pupil detection in realworld scenarios has proven to be an intricate challenge due to fast illumination changes, pupil occlusion, non-centered and off-axis eye recording, as well as physiological eye characteristics. In this paper, we approach this challenge through: I) a convolutional neural network (CNN) running in real time on a single core, II) a novel computational intensive two stage CNN for accuracy improvement, and III) a fast propability distribution based refinement method as a practical alternative to II. We evaluate the proposed approaches against the state-of-the-art pupil detection algorithms, improving the detection rate up to ~9% percent points on average over all data sets (~7% on one CPU core 7ms). This evaluation was performed on over 135,000 images: 94,000 images from the literature, and 41,000 new hand-labeled and challenging images contributed by this work (v1.0).