YOLO-ASC: You Only Look Once And See Contours

Hurtík, Petr; Molek, Vojtěch; Vlašánek, Pavel

doi:10.1109/ijcnn48605.2020.9207223

Cited by 10 publications

(10 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This means that, for example, the corners of the detections rarely align completely with any proposed homography. This could be mitigated by using a product proposal generator that returns the contours of the object instead of just an encompassing rectangle; something along the lines of the approach of Hurtik et al (2020). Utilizing contours might allow the whole detection to align with the homography, resulting in less false positives and -hopefully -a better homography.…”

Section: Using a Contour-detecting Product Proposal Generatormentioning

confidence: 99%

Computer Vision Based Planogram Compliance Evaluation

Laitala¹,

Ruotsalainen

2022

SSRN Journal

View full text Add to dashboard Cite

Section: Using a Contour-detecting Product Proposal Generatormentioning

confidence: 99%

Computer Vision Based Planogram Compliance Evaluation

Laitala¹,

Ruotsalainen

2022

SSRN Journal

View full text Add to dashboard Cite

“…The disadvantage of this approach is that for the objects of complex shapes, the bounding box also includes background, which can occupy a significant part of the area as the bounding box does not wrap the object tightly. Such behavior can decrease the performance of a classifier applied over the bounding box [1] or may not fulfill requirements of precise detection [2]. To avoid the problem, classical detectors such as Faster R-CNN [3] or RetinaNet [4] were modified into a version of Mask R-CNN [5] or RetinaMask [6].…”

Section: Problem Statementmentioning

confidence: 99%

“…Here, we show how to extend YOLO with masking functionality (instance segmentation) without a big negative impact on its speed. In our previous work [1], we were focusing on more precise detection of YOLO by means of irregular quadrangular detection. We proved that the extension for quadrangular detection converges faster.…”

Section: Instance Segmentation With Poly-yolomentioning

confidence: 99%

“…In the case of Poly-YOLO, the number of convolutional filters in the output layer has to be updated. When we detect only bounding boxes, the last layer is represented by n = n a (n c + 5) convolutional filters with a kernel of dimension (1,1), where n a is a number of anchors (nine, in our case) and n c stands for a number of classes. After integrating the extension for polygon-based object detection, we obtain n = n a (n c + 5 + 3n v ), where n v is a maximal number of detected vertices per polygon.…”

Section: Integration With Poly-yolomentioning

confidence: 99%

See 1 more Smart Citation

Poly-YOLO: higher speed, more precise detection and instance segmentation for YOLOv3

Hurtík¹,

Molek²,

Hůla³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

We present a new version of YOLO with better performance and extended with instance segmentation called Poly-YOLO. Poly-YOLO builds on the original ideas of YOLOv3 and removes two of its weaknesses: a large amount of rewritten labels and inefficient distribution of anchors. Poly-YOLO reduces the issues by aggregating features from a light SE-Darknet-53 backbone with a hypercolumn technique, using stairstep upsampling, and produces a single scale output with high resolution. In comparison with YOLOv3, Poly-YOLO has only 60% of its trainable parameters but improves mAP by a relative 40%. We also present Poly-YOLO lite with fewer parameters and a lower output resolution. It has the same precision as YOLOv3, but it is three times smaller and twice as fast, thus suitable for embedded devices. Finally, Poly-YOLO performs instance segmentation using bounding polygons. The network is trained to detect size-independent polygons defined on a polar grid. Vertices of each polygon are being predicted with their confidence, and therefore Poly-YOLO produces polygons with a varying number of vertices. Source code is available at https://gitlab.com/irafm-ai/poly-yolo.

show abstract

“…But after the pandemic happened, since 2020, research work on mask detection based on machine learning algorithms has skyrocketed. The YOLO-ASC approach was proposed which allows the detection of objects more accurately even without a background in real-time [9]. YOLO could also be improved in terms of predicting the absolute distance of objects using only information from a monocular camera with an inference speed of 45 frames per second [10].…”

Section: Introductionmentioning

confidence: 99%

An Automated Face-mask Detection System using YOLOv5 for Preventing Spread of COVID-19

Jishan¹,

Alam²,

Mazumder³

et al. 2023

IJC

View full text Add to dashboard Cite

Object detection systems based on deep learning have been immensely successful incomplex object detection tasks images and have shown potential in a wide range of real-life applicationsincluding the COVID-19 pandemic. One of the key challenges in containing and mitigating the infectionamong the population is to ensure and enforce the proper use of face masks. The objective of this paperis to detect the proper use of facial masks among the urban population in a megacity. In this study, wetrained and validated a new dataset to detect images such as ‘with mask’, ‘without mask’, and ‘masknot in position’ using YOLOv5. The dataset is comprised of 6550 images with the three classes. Thedataset attained a commendable performance accuracy of 95% on mAP. This study can be implementedfor automated scanning for monitoring the proper use of face masks in different settings of public spaces.

show abstract

YOLO-ASC: You Only Look Once And See Contours

Cited by 10 publications

References 21 publications

Computer Vision Based Planogram Compliance Evaluation

Computer Vision Based Planogram Compliance Evaluation

Poly-YOLO: higher speed, more precise detection and instance segmentation for YOLOv3

An Automated Face-mask Detection System using YOLOv5 for Preventing Spread of COVID-19

Contact Info

Product

Resources

About