Robust object detection methods in traffic surveillance scenarios often encounters challenges due to large-scale deformations and illumination variations in outdoor scenes. To enhance the tolerance of such methods against these variations, we design a cross-scale and illumination-invariant detection model (CSIM) based on the You Only Look Once (YOLO) architecture. A main cause of false detection in large-scale detection tasks is the inconsistency between various feature scales. To address this issue, we introduce an adaptive cross-scale feature fusion model to ensure the consistency of the constructed feature pyramid. To overcome the influence of uneven light, we build an illumination-invariant chromaticity space on the CSIM model, which is independent of the correlated color temperature. In addition, we adopt spatial attention modules, K-means clustering and the Mish activation function for further model optimization. The obtained experimental results show that the proposed CSIM produces excellent detection results for addressing the challenges derived from large-scale deformations and the illumination changes encountered during traffic surveillance. Compared with state-of-the-art object detection methods on public datasets, our proposed model has achieved competitive results in robust object detection tasks in traffic surveillance scenarios.

Spiking Neural Networks (SNNs) have attracted significant attention for their energy-efficient and brain-inspired event-driven properties. Recent advancements, notably Spiking-YOLO, have enabled SNNs to undertake advanced object detection tasks. Nevertheless, these methods often suffer from increased latency and diminished detection accuracy, rendering them less suitable for latency-sensitive mobile platforms. Additionally, the conversion of artificial neural networks (ANNs) to SNNs frequently compromises the integrity of the ANNs’ structure, resulting in poor feature representation and heightened conversion errors. To address the issues of high latency and low detection accuracy, we introduce two solutions: timestep compression and spike-time-dependent integrated (STDI) coding. Timestep compression effectively reduces the number of timesteps required in the ANN-to-SNN conversion by condensing information. The STDI coding employs a time-varying threshold to augment information capacity. Furthermore, we have developed an SNN-based spatial pyramid pooling (SPP) structure, optimized to preserve the network’s structural efficacy during conversion. Utilizing these approaches, we present the ultralow latency and highly accurate object detection model, SUHD. SUHD exhibits exceptional performance on challenging datasets like PASCAL VOC and MS COCO, achieving a remarkable reduction of approximately 750 times in timesteps and a 30% enhancement in mean average precision (mAP) compared to Spiking-YOLO on MS COCO. To the best of our knowledge, SUHD is currently the deepest spike-based object detection model, achieving ultralow timesteps for lossless conversion.

End-to-end autonomous driving provides a feasible way to automatically maximize overall driving system performance by directly mapping the raw pixels from a front-facing camera to control signals. Recent advanced methods construct a latent world model to map the high dimensional observations into compact latent space. However, the latent states embedded by the world model proposed in previous works may contain a large amount of task-irrelevant information, resulting in low sampling efficiency and poor robustness to input perturbations. Meanwhile, the training data distribution is usually unbalanced, and the learned policy is challenging to cope with the corner cases during the driving process. To solve the above challenges, we present a SEM antic M asked recurrent world model ( SEM2 ), which introduces a semantic filter to extract key driving-relevant features and make decisions via the filtered features, and is trained with a multi-source data sampler, which aggregates common data and multiple corner case data in a single batch, to balance the data distribution. Extensive experiments on CARLA show our method outperforms the state-of-the-art approaches in terms of sample efficiency and robustness to input permutations.

The spiking neural networks (SNNs), as brain-inspired neural networks, have received noteworthy attention due to their advantages of low power consumption, high parallelism, and high fault tolerance. While SNNs have shown promising results in uni-modal data tasks, their deployment in multi-modal audiovisual classification remains limited, and the effectiveness of capturing correlations between visual and audio modalities in SNNs needs improvement. To address these challenges, we propose a novel model called Spiking Multi-Model Transformer (SMMT) that combines SNNs and Transformers for multi-modal audiovisual classification. The SMMT model integrates uni-modal sub-networks for visual and auditory modalities with a novel Spiking Cross-Attention module for fusion, enhancing the correlation between visual and audio modalities. This approach leads to competitive accuracy in multi-modal classification tasks with low energy consumption, making it an effective and energy-efficient solution. Extensive experiments on a public event-based dataset(N-TIDIGIT&MNIST-DVS) and two self-made audiovisual datasets of real-world objects(CIFAR10-AV and UrbanSound8K-AV) demonstrate the effectiveness and energy efficiency of the proposed SMMT model in multi-modal audio-visual classification tasks. Our constructed multi-modal audiovisual datasets can be accessed at https://github.com/Guo-Lingyue/SMMT.

Due to the high costs and difficulties to make the short-current experiment of large power transformers, it is critical to do the electromagnetic force analysis of the transformer windings through finite element software. Research in this paper establishes the analysis model of a 50000 KVA/110 KV three phase power transformer. In previous studies, short-circuit impedance is seldom evaluated for its effect to the accumulated effect of the electromagnetic force. A novel approach with considering of the short-circuit impedance is proposed to simulate the accumulated deformation. Short-circuit current is applied to this model, testing the electromagnetic forces of the transformer windings. Results of the research is significant to the evaluation of winding deformation state.

This paper presents an intelligent inspection system for power insulators, integrating Unmanned Aerial Vehicles (UAVs) to address complex conditions. A novel image enhancement technique, based on the multi-scale Retinex algorithm with combined illumination correction and compensation, is proposed to enhance both the overall image quality of insulators and the visual perception for human observers. The study employs the state-of-the-art YOLOv7 target detector for automated identification of insulator defects. The YOLOv7 model undergoes further optimization, enhancing the feature extraction structure, incorporating the SE attention mechanism, and refining the loss function. These modifications enable the model to adapt to more intricate intelligent inspection tasks related to transmission line insulators, resulting in improved efficiency and detection accuracy. Building on this research out-comes, our team introduces an intelligent power insulator detection software system tailored for UAV inspection scenarios