This dataset was developed to support an assistive belt system designed to help visually impaired individuals navigate outdoor environments safely. It contains annotated images covering 15 object classes relevant to outdoor navigation: person, car, bus, truck, bicycle, motorcycle, tree, trash (bin), pedestrian crossing signal (green light), pedestrian crossing signal (red light), crosswalk, door, stairs, stop sign, and bench. The dataset combines images from open-source datasets with manually collected and annotated images. All annotations are provided in YOLO format (bounding boxes with class labels), making the dataset ready for training object detection models (e.g., YOLOv5/v8) for real-time outdoor obstacle and landmark detection in assistive navigation applications. Intended use cases include training and benchmarking computer vision models for wearable/embedded assistive technology aimed at improving outdoor mobility and safety for visually impaired users. This dataset incorporates images and annotations derived from the COCO dataset (CC BY 4.0) and the Open Images dataset (CC BY 4.0), in addition to manually collected and annotated images.
The use of Advanced Driver Assistance Systems (ADAS) heavily depends on the perception models to make real-time decisions but the traditional methods have tended to use specific confidence thresholds to make the trade-offs between missed detections and false alarms to be not opti…
# Overview The **V2X Collision Avoidance System** is an intelligent transportation platform that integrates **Vehicle-to-Everything (V2X)** communication, **embedded systems**, **artificial intelligence**, and **sensor fusion** to improve road safety through real-time collision p…
Abstract: Deepfake technology, driven by generative models such as GANs and diffusion architectures, has enabled the creation of highly realistic manipulated media capable of deceiving both visual and auditory perception. Such forgeries pose significant risks to identity verifica…
D013 Prime Elementology — Prime Spectral Descriptor Atlas for Synthetic RF Dataset ID: D013Version: 2.0Dataset Type: Synthetic Research DatasetAuthor: Phan Thành TrungORCID: 0009-0000-7520-6781DOI: 10.5281/zenodo.21569013 1. Overview D013 Prime Elementology — Prime Spectral Descr…
This repository contains the official open-source code for [YOLO vs. Diffusion Networks for Underground PipeDetection: A Case Study Using Ground PenetratingRadar Data]. ### OverviewThis software provides an end-to-end implementation of deep Learning for Pipeline Detection Using G…
🌟 Summary Huawei Ascend support arrives in Ultralytics, enabling YOLO models to export and run as hardware-optimized .om models on Ascend NPUs. 🚀 📊 Key Changes Huawei Ascend export and inference 🧠 Adds format=ascend to compile YOLO models through Huawei's CANN ATC compiler. P…