BrainCog: A spiking neural network based, brain-inspired cognitive intelligence engine for brain-inspired AI and brain simulation.

Spiking neural networks (SNNs) serve as a promising computational framework for integrating insights from the brain into artificial intelligence (AI). Existing software infrastructures based on SNNs exclusively support brain simulation or brain-inspired AI, but not both simultaneously. To decode the nature of biological intelligence and create AI, we present the brain-inspired cognitive intelligence engine (BrainCog). This SNN-based platform provides essential infrastructure support for developing brain-inspired AI and brain simulation. BrainCog integrates different biological neurons, encoding strategies, learning rules, brain areas, and hardware-software co-design as essential components. Leveraging these user-friendly components, BrainCog incorporates various cognitive functions, including perception and learning, decision-making, knowledge representation and reasoning, motor control, social cognition, and brain structure and function simulations across multiple scales. BORN is an AI engine developed by BrainCog, showcasing seamless integration of BrainCog's components and cognitive functions to build advanced AI models and applications.

Learning Semantic-Aware Local Features for Long Term Visual Localization.

Extracting robust and discriminative local features from images plays a vital role for long term visual localization, whose challenges are mainly caused by the severe appearance differences between matching images due to the day-night illuminations, seasonal changes, and human activities. Existing solutions resort to jointly learning both keypoints and their descriptors in an end-to-end manner, leveraged on large number of annotations of point correspondence which are harvested from the structure from motion and depth estimation algorithms. While these methods show improved performance over non-deep methods or those two-stage deep methods, i.e., detection and then description, they are still struggled to conquer the problems encountered in long term visual localization. Since the intrinsic semantics are invariant to the local appearance changes, this paper proposes to learn semantic-aware local features in order to improve robustness of local feature matching for long term localization. Based on a state of the art CNN architecture for local feature learning, i.e., ASLFeat, this paper leverages on the semantic information from an off-the-shelf semantic segmentation network to learn semantic-aware feature maps. The learned correspondence-aware feature descriptors and semantic features are then merged to form the final feature descriptors, for which the improved feature matching ability has been observed in experiments. In addition, the learned semantics embedded in the features can be further used to filter out noisy keypoints, leading to additional accuracy improvement and faster matching speed. Experiments on two popular long term visual localization benchmarks (Aachen Day and Night v1.1, Robotcar Seasons) and one challenging indoor benchmark (InLoc) demonstrate encouraging improvements of the localization accuracy over its counterpart and other competitive methods.

A Performance Evaluation of Local Features for Image-Based 3D Reconstruction.

This paper performs a comprehensive and comparative evaluation of the state-of-the-art local features for the task of image-based 3D reconstruction. The evaluated local features cover the recently developed ones by using powerful machine learning techniques and the elaborately designed handcrafted features. To obtain a comprehensive evaluation, we choose to include both float type features and binary ones. Meanwhile, two kinds of datasets have been used in this evaluation. One is a dataset of many different scene types with groundtruth 3D points, containing images of different scenes captured at fixed positions, for quantitative performance evaluation of different local features in the controlled image capturing situation. The other dataset contains Internet scale image sets of several landmarks with a lot of unrelated images, which is used for qualitative performance evaluation of different local features in the free image collection situation. Our experimental results show that binary features are competent to reconstruct scenes from controlled image sequences with only a fraction of processing time compared to using float type features. However, for the case of a large scale image set with many distracting images, float type features show a clear advantage over binary ones. Currently, the most traditional SIFT is very stable with regard to scene types in this specific task and produces very competitive reconstruction results among all the evaluated local features. Meanwhile, although the learned binary features are not as competitive as the handcrafted ones, learning float type features with CNN is promising but still requires much effort in the future.

Efficient coding matters in the organization of the early visual system.

Individual areas in the brain are organized into a hierarchical network as a result of evolution. Previous work indicated that the receptive fields (RFs) of individual areas have been evolved to favor metabolically efficient neural codes. In this paper, we propose that not only the RFs of individual areas, but also the organization of adjacent neurons and the hierarchical structure composed of these areas have been evolved to support efficient coding. To verify this hypothesis, we introduce a feed-forward three-layer network to simulate the early stages of human visual system. We emphasize that the network is not a purely feed-forward one since it also includes intra-layer connections, which are essential but usually ignored in the literature. Simulation results strongly reveal that (1) the obtained RFs of the simulated retinal ganglion cells (RGCs) or neurons in the lateral geniculate nucleus (LGN) and V1 simple neurons are consistent to the neurophysiological data; (2) the responses of closer RGCs are more correlated, and V1 simple neurons with similar orientations prefer to cluster together; (3) the hierarchical organization of the early visual system is beneficial for saving energy, which accords with the requirement of metabolically efficient neural coding in the process of human brain evolution.

A Fast Contour Detection Model Inspired by Biological Mechanisms in Primary Vision System.

Compared to computer vision systems, the human visual system is more fast and accurate. It is well accepted that V1 neurons can well encode contour information. There are plenty of computational models about contour detection based on the mechanism of the V1 neurons. Multiple-cue inhibition operator is one well-known model, which is based on the mechanism of V1 neurons' non-classical receptive fields. However, this model is time-consuming and noisy. To solve these two problems, we propose an improved model which integrates some additional other mechanisms of the primary vision system. Firstly, based on the knowledge that the salient contours only occupy a small portion of the whole image, the prior filtering is introduced to decrease the running time. Secondly, based on the physiological finding that nearby neurons often have highly correlated responses and thus include redundant information, we adopt the uniform samplings to speed up the algorithm. Thirdly, sparse coding is introduced to suppress the unwanted noises. Finally, to validate the performance, we test it on Berkeley Segmentation Data Set. The results show that the improved model can decrease running time as well as keep the accuracy of the contour detection.

Receptive fields selection for binary feature description.

Feature description for local image patch is widely used in computer vision. While the conventional way to design local descriptor is based on expert experience and knowledge, learning-based methods for designing local descriptor become more and more popular because of their good performance and data-driven property. This paper proposes a novel data-driven method for designing binary feature descriptor, which we call receptive fields descriptor (RFD). Technically, RFD is constructed by thresholding responses of a set of receptive fields, which are selected from a large number of candidates according to their distinctiveness and correlations in a greedy way. Using two different kinds of receptive fields (namely rectangular pooling area and Gaussian pooling area) for selection, we obtain two binary descriptors RFDR and RFDG .accordingly. Image matching experiments on the well-known patch data set and Oxford data set demonstrate that RFD significantly outperforms the state-of-the-art binary descriptors, and is comparable with the best float-valued descriptors at a fraction of processing time. Finally, experiments on object recognition tasks confirm that both RFDR and RFDG successfully bridge the performance gap between binary descriptors and their floating-point competitors.