Search Results (6)

Insect bites are a common cause of skin irritation and can contribute to disease transmission through vector-borne pathogens. Early identification of the likely biting organism can assist preliminary guidance (e.g., monitoring for warning signs, considering exposure history) and may reduce complications through timely follow-up. This paper studies a compact attention-guided learning framework for multiclass insect-bite image classification under strict storage constraints. A teacher network (BiteAI-T) based on MobileNetV3-Small is trained with spatial attention pooling to emphasize lesion-relevant regions while maintaining an efficient backbone. A lightweight depthwise-separable student (BiteAI-S) is trained using multi-level knowledge distillation that combines softened-logit matching with intermediate supervision through attention-map alignment and pooled-feature matching. Model storage is further reduced through weight-only quantization-aware training using an LSQ-inspired learnable scaling factor; BatchNorm running statistics are frozen during quantization fine-tuning to improve stability. Experiments on an eight-class dataset (ants, bed bugs, chiggers, fleas, mosquitos, no bites, spiders, ticks) show that BiteAI-T reaches 93.75% test accuracy. For deployment, we export (i) a TorchScript Lite teacher artifact (BiteAI-T_Lite, 2.35 MB) and (ii) a weight-only int8 student artifact (BiteAI-S_int8, 0.992 MB). Comparative results are also reported for an SVD-compressed + fine-tuned FP16 variant (92.66% test accuracy, 2.84 MB), illustrating accuracy–size trade-offs across compression strategies. Full article

Edge AI is redefining the deployment of computer vision systems by enabling real-time inference directly on resource-constrained edge devices. This shift offers significant advantages in terms of reduced latency, data privacy, and operational autonomy in bandwidth-limited computing environments. In this paper, we present a systematic performance benchmarking of multiple variants of YOLOv8 on the NVIDIA Jetson Orin NX platform, focusing on object detection tasks. We evaluate inference latency, frame throughput, and computational resource usage across varying input sizes and model complexities. Furthermore, we validate the deployment effectiveness through practical use cases, such as vehicle and package detection. Our findings show that the TensorRT model outperforms PyTorch by 17.7% at a batch size of 2, although PyTorch presents greater stability at larger batch sizes (e.g., 8), where TensorRT encounters resource constraints. In terms of memory usage, it increases linearly with batch size: 69% from batch 1 to 4, with TensorRT requiring 429.20 MB at batch size 2 compared to PyTorch’s 451.24 MB. Furthermore, the processing time per image decreases by 42% when scaling from batch size 1 to 4, highlighting a critical saturation point for edge resources. In summary, the results provide insight into the trade-offs between model size and speed, offering guidance for selecting detection architectures tailored to real-time edge applications. Full article

Automated anomaly detection in transportation infrastructure is essential for enhancing safety and reducing the operational costs associated with manual inspection protocols. This study presents an improved neuromorphic vision system, which extends the prior SIFT-SNN (scale-invariant feature transform–spiking neural network) proof-of-concept by incorporating temporal feature aggregation for context-aware and sequence-stable detection. Analysis of classical stitching-based pipelines exposed sensitivity to motion and lighting variations, motivating the proposed temporally smoothed neuromorphic design. SIFT keypoints are encoded into latency-based spike trains and classified using a leaky integrate-and-fire (LIF) spiking neural network implemented in PyTorch. Evaluated across three hardware configurations—an NVIDIA RTX 4060 GPU, an Intel i7 CPU, and a simulated Jetson Nano—the system achieved 92.3% accuracy and a macro F1 score of 91.0% under five-fold cross-validation. Inference latencies were measured at 9.5 ms, 26.1 ms, and ~48.3 ms per frame, respectively. Memory footprints were under 290 MB, and power consumption was estimated to be between 5 and 65 W. The classifier distinguishes between safe, partially dislodged, and fully dislodged barrier pins, which are critical failure modes for the Auckland Harbour Bridge’s Movable Concrete Barrier (MCB) system. Temporal smoothing further improves recall for ambiguous cases. By achieving a compact model size (2.9 MB), low-latency inference, and minimal power demands, the proposed framework offers a deployable, interpretable, and energy-efficient alternative to conventional CNN-based inspection tools. Future work will focus on exploring the generalisability and transferability of the work presented, additional input sources, and human–computer interaction paradigms for various deployment infrastructures and advancements. Full article

Plasma spray coating employs a high-temperature plasma jet to melt and deposit powdered materials onto substrates and plays a critical role in aerospace and manufacturing. Despite its importance, the influence of torch behavior, particularly the thermal response of plasma to gas composition changes, remains inadequately characterized. In this study, a three-dimensional MHD simulation using OpenFOAM (v2112) was performed on a Metco 9MB plasma torch operating in an Ar–H₂–N₂ environment under the LTE assumption to investigate the effect of nitrogen addition. The simulation revealed that increasing nitrogen levels results in a dual effect on the temperature distribution: temperatures rise near the cathode tip and decrease downstream, likely due to variations in the net emission coefficient and enthalpy characteristics of nitrogen. Furthermore, although the outlet velocity remained largely unaffected, the Mach number increased as the nitrogen reduced the speed of sound. These findings provide essential insights for optimizing ternary gas mixtures to enhance coating efficiency in thermal spray applications. Full article

Experimental evidence of swirling effects in 3D trajectories of in-flight particles is presented based on static and dynamic footprints analysis as a function of stand-off distance of Al₂O₃ deposited employing a Metco-9MB torch. Swirling effects were validated with a proprietary computational fluid dynamics (CFD) code that considers an argon-hydrogen plasma stream, in-flight particles trajectories, both creating the spray cone, and particle impact to form a footprint on a fixed substrate located at different distances up to 120 mm. Static and dynamic footprints showed that swirl produces a slight deviation of individual particle trajectories and thus footprint rotation, which may affect coating characteristics. Full article

The demand for suspensions that are used in thermal spray processes is expanding from research labs using the lab-prepared suspensions toward actual coating production in different industrial sectors. Industrial applications dictate the reduced production time and effort, which may in turn justify the development of the market for ready-to-use commercial suspensions. To this end, some of the powder suppliers have already taken steps forward by introducing, to the market, suspensions of some of the most used materials, such as yttria-stabilized zirconia (YSZ), alumina, and titania. However, there is a need to compare the suspension characteristics over time and the resultant coatings when using these suspensions when compared with the freshly prepared homemade suspensions. In this work, such a comparison is done using YSZ suspensions of the sub-micron to a few micron powders. In addition, some changes in the suspensions’ formula were performed as a tool to vary the coatings’ microstructures in a more predictable way, without any variation of the spray parameters. The coatings were generated while using both radial and axial injection of the suspensions into Oerlikon-Metco 3MB and Mettech Axial III plasma spray torches, respectively. A clear effect of suspension viscosity on the coating microstructure was observed using the 3MB torch with a radial injection of suspension (i.e., cross flow atomization). However, the viscosity role was not dominant when using the Axial III torch with an axial feed injection system (i.e., coaxial flow atomization). Full article