Accurate and automated detection of surface knots on sawn timbers using YOLO-V5 model

Fang, Yiming; Guo, Xianxin; Chen, Kun; Zhou, Zhu; Ye, Qing

doi:10.15376/biores.16.3.5390-5406

Cited by 120 publications

(53 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The knottiness data used here were produced by the company Finnos using their in-house algorithms-we did not know their exact methods but rely on their credibility as trusted industrial supplier of the X-raying technology. In general, knot identification from Xrayed log tomographic images in the sawmilling industry is mostly based on convolutional neural networks locally trained for object detection (Fang, et al 2021). Ring properties were measured from the X-ray densitometric images, and due to the resolution of the images, the measurements of the small rings may contain relatively more error, than those of the larger rings.…”

Section: Analysis Of the Resultsmentioning

confidence: 99%

How tree morphology reflects ring properties in Norway spruce?

Pehkonen¹,

Holopainen²,

Kukko³

et al. 2022

Preprint

View full text Add to dashboard Cite

We studied the relationships of wood anatomy (ring size and composition) with select morphological features (stem geometry and branching) in Norway spruce (Picea Abies H. Karst). We acquired the morphological modelling data by tracking the 1179 sawlogs of 479 trees from 14 even-aged, spruce-dominated stands in South-Eastern Finland to a sawmill, where X-ray log tomographic measurements were taken. We smoothed three measures of knottiness with continuous functions of log height and diameter, namely knot index (ratio of knot volume to log volume), knot size index (ratio of knot volume to number of whorls) and shoot lengths (whorl-to-whorl distances). The ring properties data came from a subset of 172 ring sample discs from variable heights in 51 of these trees. X-ray densitometric measurements were conducted to define the cambial age (CA), ring area (RA), latewood percentage (LWP), and ring density (RD) of each ring. We established reference mixed models for the ring properties like those frequently used in the literature. The reference models explained up to 80% of CA and RA, and around 44% of LWP and RD, with additional 6–20% explained by stand-level random variance. Next, we tested the relationships of the ring properties with the stem and knottiness features at individual ring-level. The tree morphology showed clear connection with the CA and RA: stem size, relative height, knottiness, within-crown positions, and apical shoot lengths explained 69% of CA, and 53% of RA, in addition to 12–16% stand-level variance. However, the tree morphology accounted for only 7% and 4% of the variation in LWP and RD, respectively, while the stand-level random variance accounted for another 22%–23% in these models. Based on previous literature, LWP and RD are largely affected by site fertility and annual weather conditions that control especially the timing and rate of early- and latewood production, and cell-wall thickness in latewood. Therefore, the statistical predictions of such subtle, intra-annual changes with the coarser morphological developments of crown and stem remain limited in accuracy. The analysis of our results revealed that increased crown vigour and longevity was reflected in increased earlywood production in juvenile wood. Therefore, the higher the growth rates in young trees, the higher the percentage of low-density juvenile wood in mature stems, especially on fertile and moist sites where trees favoured water uptake in the expense of structural support. We found limited possibilities to increase RD on the more fertile sites by increasing stocking density. The stem and knottiness features used here resemble those re-constructible from laser-scanners and photogrammetric sensors increasingly often used in both forested and industrial environments. The results here highlighted some of the possibilities and remaining knowledge gaps regarding the relationships of observable tree morphologies and the wood anatomy that require solving before feasible methods to, e.g., remote-sensing-aided wood property surveys are possible.

show abstract

Section: Analysis Of the Resultsmentioning

confidence: 99%

How tree morphology reflects ring properties in Norway spruce?

Pehkonen¹,

Holopainen²,

Kukko³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Yolo has been updated to version five and is regarded as the state-of-the-art algorithm for object detection ( 24 ). It has been applied in many daily life aspects, such as the detection of surface knots ( 25 ) and real-time vehicles ( 26 ), as well as in various medical fields, including face mask recognition ( 27 ), breast tumor detection and classification ( 28 ), and chest abnormality detection ( 29 ). This study showed that the basic deep learning model Yolo V5 could handle the cyst-detection task, attaining F1, precision, and mAP scores of 0.832, 0.843, and 0.821, respectively.…”

Section: Discussionmentioning

confidence: 99%

Automated detection of knee cystic lesions on magnetic resonance imaging using deep learning

Tang

Xianyue

et al. 2022

Front. Med.

View full text Add to dashboard Cite

BackgroundCystic lesions are frequently observed in knee joint diseases and are usually associated with joint pain, degenerative disorders, or acute injury. Magnetic resonance imaging-based, artificial intelligence-assisted cyst detection is an effective method to improve the whole knee joint analysis. However, few studies have investigated this method. This study is the first attempt at auto-detection of knee cysts based on deep learning methods.MethodsThis retrospective study collected data from 282 subjects with knee cysts confirmed at our institution from January to October 2021. A Squeeze-and-Excitation (SE) inception attention-based You only look once version 5 (SE-YOLOv5) model was developed based on a self-attention mechanism for knee cyst-like lesion detection and differentiation from knee effusions, both characterized by high T2-weighted signals in magnetic resonance imaging (MRI) scans. Model performance was evaluated via metrics including accuracy, precision, recall, mean average precision (mAP), F1 score, and frames per second (fps).ResultsThe deep learning model could accurately identify knee MRI scans and auto-detect both obvious cyst lesions and small ones with inconspicuous contrasts. The SE-YOLO V5 model constructed in this study yielded superior performance (F1 = 0.879, precision = 0.887, recall = 0.872, all class mAP0.5 = 0.944, effusion mAP = 0.945, cyst mAP = 0.942) and improved detection speed compared to a traditional YOLO model.ConclusionThis proof-of-concept study examined whether deep learning models could detect knee cysts and distinguish them from knee effusions. The results demonstrated that the classical Yolo V5 and proposed SE-Yolo V5 models could accurately identify cysts.

show abstract

“…YOLO outperforms the other traditional object detection algorithms by examining (1) the images only once to find the objects inside them. Among all previous versions, YOLO V5 has a faster processing speed with the most advanced structure using parallel calculations [42,43]. Many new techniques are used in YOLO V5, such backbone (CSP-Darknet), neck (PAnet), head (YOLO layer) [44].…”

Section: You Only Look Once (Yolo) V5mentioning

confidence: 99%

Classified 3D mapping and deep learning-aided signal power estimation architecture for the deployment of wireless communication systems

Egi

Eyceyurt

2022

J Wireless Com Network

View full text Add to dashboard Cite

The traditional wireless communication systems deployment models require expensive and time-consuming procedures, including environment selection (rural, urban, and suburban), drive test data collection, and analysis of the raw data. These procedures mainly utilize stochastic and deterministic approaches for signal strength prediction to locate the optimum cellular tower (eNodeB) position for 4G and 5G systems. Since environment selection is limited by urban, suburban, and rural areas, they do not cover complex macro and micro variations, especially buildings and tree canopies having a higher impact on signal fading due to scattering and absorption. Therefore, they usually end up with high prediction errors. This article proposes an efficient architecture for the deployment of communication systems. The proposed method determines the effect of the environment via extracting tree and building properties by using a classified 3D map and You Only Look Once (YOLO) V5, which is one of the most efficient deep learning algorithms. According to the results, the mean average precision (mAP) 0.5% and mAP 0.95% accuracies are obtained as 0.96 and 0.45, and image color classification (ICC) findings indicate 77.6% accuracy on vegetation detection, especially for tree canopies. Thus, the obtained results significantly improved signal strength prediction with a 3.96% Mean Absolute Percentage Error (MAPE) rate, while other empirical models’ prediction errors fall in the range of 6.07–15.26%.

show abstract

Accurate and automated detection of surface knots on sawn timbers using YOLO-V5 model

Cited by 120 publications

References 39 publications

How tree morphology reflects ring properties in Norway spruce?

How tree morphology reflects ring properties in Norway spruce?

Automated detection of knee cystic lesions on magnetic resonance imaging using deep learning

Classified 3D mapping and deep learning-aided signal power estimation architecture for the deployment of wireless communication systems

Contact Info

Product

Resources

About