China
Africa
Explore chapters and articles related to this topic
Deep Learning in Smart Agriculture Applications
Published in Rashmi Priyadarshini, R M Mehra, Amit Sehgal, Prabhu Jyot Singh, Artificial Intelligence, 2023
For yield prediction and robot harvesting, fruit counting is one of several important factors; we cannot produce satisfactory results through traditional counting or video or camera image counting, and also these processes are time consuming. Pre-processing of those sorts of images is challenging due to occlusion and illumination. [22] has introduced the technique to spot the livestock animal like pig face recognition through the CNN technique. Conventionally, frequency identification tags were used for detecting the animals, which was a cumbersome job. To accompany a fully convolutional network, a method known as blob detection is proposed. The first step is to gather the human formed labels from the set of fruit images and then this model is trained for an image segmentation performance. Then CNN was trained to consider the bifurcated pictures and generate a middle approximation of the fruit count. The concluding stage of the work is to train a regression equation to map intermediate fruit count estimation to final human generated label count. Accuracy as well as efficiency is increased by combining deep learning with blob detection.
Machine Vision in Industry 4.0
Published in Roshani Raut, Salah-ddine Krit, Prasenjit Chatterjee, Machine Vision for Industry 4.0, 2022
Pramod Kumar, Dharmendra Singh, Jaiprakash Bhamu
Frustaci et al. (2020) designed and developed embedded machine-based system for inline geometric inspection of catalytic converter assembly process for flexible, low cost and precise inspection. Afterward, defects in geometry due to rotational shifts of interfaces about their mean positions can also be detected. Real-time counting of manually assembled components is carried out by using versatile algorithm of machine vision (Pierleoni et al., 2020). This system has capability to deal with human interactions and counting of assembled pieces by using real-time video inputs. It consists of inter-frame analysis, image preprocessing, binarisation, morphological operations and blob detection. One custom detector based on machine learning is considered a reference for comparison.
Machine learning and deep learning in agriculture
Published in Govind Singh Patel, Amrita Rai, Nripendra Narayan Das, R. P. Singh, Smart Agriculture, 2021
To accompany a fully convolution network, a method known as blob detection was proposed. The first step is to gather the human formed labels from a set of fruit images and then this model is trained for an image segmentation performance. Then CNN is used to count the bifurcated pictures and give an approximation of number of fruits. The last stage of the work involves applying a regression equation to map intermediate fruit count estimation to final human generated label count. Accuracy as well as efficiency is increased by combining deep learning with blob detection.
Phase image-guided adaptive rotation-invariant feature point detector
Published in The Imaging Science Journal, 2023
Ahmed S. Mashaly, Tarek A. Mahmoud
Conceptually, Blob detector suffers from image noise sensitivity and image texture variation because it depends on second order image derivatives. As a result, several false feature points appear at strong edge locations. Therefore, it is recommended to apply an image enhancement step before using Blob detector [7–9].
Activity representation by SURF-based templates
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2018
Md. Atiqur Rahman Ahad, JK Tan, H Kim, S Ishikawa
In this paper, we exploit the local interest points to have motion cues. These corner or interest points are very important cues for recent various methods related to action representation. Based on interest points, several smart local spatio-temporal interest point (STIP)-based approaches are proposed with significantly good performances. For example, Harris detector, Smallest Univalue Segment Assimilating Nucleus (SUSAN) detector (Smith & Brady 1997), Level curve curvature, Hessian-Laplace detector, Maximally Stable Extremal Region (MSER) detector (Matas et al. 2002), Trajkovic and Hedley corner detector (Trajković & Hedley 1998), Accelerated Segment Test-based feature detectors, Wang and Brady corner detector (Wang & Brady 1995), Harris-Laplace detector, Hessian detector (Lindeberg 1998), Difference of Gaussians (DoG) (Lowe 2004), Laplacian of Gaussian (LoG) (Lindeberg 1998; Mikolajczyk & Schmid 2004), SURF (Bay et al. 2008), Features from Accelerated Segment Test (FAST) (Rosten & Drummond 2005), etc. become prominent in different perspectives (Ahad 2011). Blob detectors are sometimes interrelated with corner detectors in some literature and used the terms interchangeably. Instead of having point-wise detection, a blob detector detects a region as a blob of circle or ellipse (Ahad 2011). Some important blob detectors are: LoG (Lindeberg 1998), DoG (Lowe 2004), MSER (Matas et al. 2002), Principal Curvature-based Region detector (Deng et al. 2007), Edge-based regions, Intensity Extrema-based Region (Mikolajczyk et al. 2005), gray-level blobs are well known. The Rotation-Invariant Feature Transform (Lazebnik et al. 2004) is a rotation-invariant generalisation of Scale-Invariant Feature Transform (SIFT) (Lowe 1999, 2004). The Principal Component Analysis SIFT (Ke & Sukthankar 2004) and Gradient Location-Orientation Histogram (Mikolajczyk & Schmid 2005), Generalized Robust Invariant Feature (Lazebnik et al. 2004) are also proposed. For recognition, few more dominant methods are Spatio-temporal interest feature points, Histogram of Oriented Flow (HOF), Histogram of Oriented Gradients (HOG) (Dalal & Triggs 2005; Laptev et al. 2008).