ultralytics/yolov5: v6.0 - YOLOv5n 'Nano' models, Roboflow integration, TensorFlow export, OpenCV DNN support

Jocher, Glenn; Stoken, Alex; Chaurasia, Ayush; Borovec, Jirka; NanoCode,; TaoXie,; Kwon, Yonghye; Michael, Kalen; Liu, Changyu; Fang, Jiacong; Abhiram,; Laughing,; tkianai,; yxNONG,; Skalski, Piotr; Hogan, A. Marie; Nadar, Jebastin; imyhxy,; Mammana, Lorenzo; AlexWang,; Fati, Cristi; Montes, Diego P.; Hájek, Jan; Diaconu, Laurentiu; Minh, Mai Thanh; Marć, Paweł; albinxavi,; fatih,; Олег,; wanghaoyang,

doi:10.5281/zenodo.5563715

Cited by 89 publications

(70 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Depending on the typologies of damage included in the training dataset, AI can assess different damages such as cracks, breakages, oil stains, etc. [102] Some recently-developed and fast-growing pre-trained models include VGG-16 [103], ResNet-v2 [104], Inceptionv4 [105], YOLOv5 [106], and EfficientNet [107]. Several variants of these models are available, including hybrid architectures such as e.g., Inception-ResNet-v2 [105].…”

Section: Traditional Enhanced and Automatic Visual Inspection (Vi)mentioning

confidence: 99%

Non-Destructive Techniques for the Condition and Structural Health Monitoring of Wind Turbines: A Literature Review of the Last 20 Years

Civera

Surace

2022

Sensors

View full text Add to dashboard Cite

A complete surveillance strategy for wind turbines requires both the condition monitoring (CM) of their mechanical components and the structural health monitoring (SHM) of their load-bearing structural elements (foundations, tower, and blades). Therefore, it spans both the civil and mechanical engineering fields. Several traditional and advanced non-destructive techniques (NDTs) have been proposed for both areas of application throughout the last years. These include visual inspection (VI), acoustic emissions (AEs), ultrasonic testing (UT), infrared thermography (IRT), radiographic testing (RT), electromagnetic testing (ET), oil monitoring, and many other methods. These NDTs can be performed by human personnel, robots, or unmanned aerial vehicles (UAVs); they can also be applied both for isolated wind turbines or systematically for whole onshore or offshore wind farms. These non-destructive approaches have been extensively reviewed here; more than 300 scientific articles, technical reports, and other documents are included in this review, encompassing all the main aspects of these survey strategies. Particular attention was dedicated to the latest developments in the last two decades (2000–2021). Highly influential research works, which received major attention from the scientific community, are highlighted and commented upon. Furthermore, for each strategy, a selection of relevant applications is reported by way of example, including newer and less developed strategies as well.

show abstract

Section: Traditional Enhanced and Automatic Visual Inspection (Vi)mentioning

confidence: 99%

Non-Destructive Techniques for the Condition and Structural Health Monitoring of Wind Turbines: A Literature Review of the Last 20 Years

Civera

Surace

2022

Sensors

View full text Add to dashboard Cite

show abstract

“…YOLO5Face Similar to TinaFace, the creators of this model treated faced detection as a generic object detection task [66]. The CSPNet used in YOLOv5 [24] in combination with the FPN [49] serves as the backbone. At the end of the backbone, a SPP block [38] block is added and smaller kernel sizes are used in comparison to YOLOv5 [24].…”

Section: Academic Modelsmentioning

confidence: 99%

“…The CSPNet used in YOLOv5 [24] in combination with the FPN [49] serves as the backbone. At the end of the backbone, a SPP block [38] block is added and smaller kernel sizes are used in comparison to YOLOv5 [24]. Also, an optional sixth PAN [50] is added to the FPN.…”

Section: Academic Modelsmentioning

confidence: 99%

Are Commercial Face Detection Models as Biased as Academic Models?

Dooley¹,

Wei²,

Goldstein³

et al. 2022

Preprint

View full text Add to dashboard Cite

As facial recognition systems are deployed more widely, scholars and activists have studied their biases and harms. Audits are commonly used to accomplish this and compare the algorithmic facial recognition systems' performance against datasets with various metadata labels about the subjects of the images. Seminal works have found discrepancies in performance by gender expression, age, perceived race, skin type, etc. These studies and audits often examine algorithms which fall into two categories: academic models or commercial models. We present a detailed comparison between academic and commercial face detection systems, specifically examining robustness to noise. We find that state-of-the-art academic face detection models exhibit demographic disparities in their noise robustness, specifically by having statistically significant decreased performance on older individuals and those who present their gender in a masculine manner. When we compare the size of these disparities to that of commercial models, we conclude that commercial models -in contrast to their relatively larger development budget and industry-level fairness commitments -are always as biased or more biased than an academic model. Recently, Dooley et al. [18] performed the first benchmark of commercial face detection systems' robustness to natural noise. They found that each of Amazon Rekognition, Microsoft Azure, and * Equal contribution.

show abstract

“…In 2020, YOLOv4 was proposed byBochkovskiy et al (Bochkovskiy et al, 2020), which utilized a new backend network called CSPDarknet53 with spatial attention and Mish-activation to improve the mAP and fps (on MS COCO Dataset) by 10% and 12%, respectively compared to YOLOv3. Within one month of YOLOv4, Jocher released the YOLOv5(Jocher et al, 2020) version, which achieved state-of-the-art performance mAP and fps. The YOLOv5x6 model reported the highest reported mAP of 55.4% on MS COCO Val 2017 dataset with an inference time of 19.4ms per image and is currently ranked top among state-of-the-art object detectors; a comparison of various models with EfficientDet (Tan et al, 2020) is shown in Figure 2.…”

mentioning

confidence: 99%

“…Figure 2. Detection accuracy and inference time comparison of YOLOv5 models withEfficientDet(Jocher et al, 2020) …”

mentioning

confidence: 99%

Remote Sensing Super-resolution Object Detection (RSSOD) Dataset

2021

View full text Add to dashboard Cite

For the past two decades, there have been significant efforts to develop methods for object detection in Remote Sensing (RS) images. In most cases, the datasets for small object detection in remote sensing images are inadequate. Many researchers used scene classification datasets for object detection, which has its limitations; for example, the large-sized objects outnumber the small objects in object categories. Thus, they lack diversity; this further affects the detection performance of small object detectors in RS images. This paper reviews current datasets and object detection methods (deep learning-based) for remote sensing images. We also propose a large-scale, publicly available benchmark Remote Sensing Super-resolution Object Detection (RSSOD) dataset. The RSSOD dataset consists of 1,759 hand-annotated images with 22,091 instances of very high resolution (VHR) images with a spatial resolution of ~0.05 m. There are five classes with varying frequencies of labels per class; the images are annotated in YOLO and COCO format. The image patches are extracted from satellite images, including real image distortions such as tangential scale distortion and skew distortion. The proposed RSSOD Preprint submitted to an Elsevier Journal for Review 2 dataset will help researchers benchmark the state-of-the-art object detection methods across various classes, especially for small objects using image super-resolution. We also propose a novel Multi-class Cyclic super-resolution Generative adversarial network with Residual feature aggregation (MCGR) and auxiliary YOLOv5 detector to benchmark image super-resolutionbased object detection and compare with the existing state-of-the-art methods based on image super-resolution (SR). The proposed MCGR achieved state-of-the-art performance for image SR with an improvement of 1.2dB PSNR compared to the current state-of-the-art NLSN method. MCGR achieved best object detection mAPs of 0.758, 0.881, 0.841, and 0.983, respectively, for five-class, four-class, two-class, and single classes, respectively surpassing the performance of the state-of-the-art object detectors YOLOv5, EfficientDet, Faster RCNN, SSD, and RetinaNet.

show abstract

ultralytics/yolov5: v6.0 - YOLOv5n 'Nano' models, Roboflow integration, TensorFlow export, OpenCV DNN support

Cited by 89 publications

References 0 publications

Non-Destructive Techniques for the Condition and Structural Health Monitoring of Wind Turbines: A Literature Review of the Last 20 Years

Non-Destructive Techniques for the Condition and Structural Health Monitoring of Wind Turbines: A Literature Review of the Last 20 Years

Are Commercial Face Detection Models as Biased as Academic Models?

Remote Sensing Super-resolution Object Detection (RSSOD) Dataset

Contact Info

Product

Resources

About