A Methodology for Utilizing Vector Space to Improve the Performance of a Dog Face Identification Model

Recent improvements in the performance of the human face recognition model have led to the development of relevant products and services. However, research in the similar field of animal face identification has remained relatively limited due to the greater diversity and complexity in shape and the lack of relevant data for animal faces such as dogs. In the face identification model using triplet loss, the length of the embedding vector is normalized by adding an L2-normalization (L2-norm) layer for using cosine-similarity-based learning. As a result, object identification depends only on the angle, and the distribution of the embedding vector is limited to the surface of a sphere with a radius of 1. This study proposes training the model from which the L2-norm layer is removed by using the triplet loss to utilize a wide vector space beyond the surface of a sphere with a radius of 1, for which a novel loss function and its two-stage learning method. The proposed method classifies the embedding vector within a space rather than on the surface, and the model’s performance is also increased. The accuracy, one-shot identification performance, and distribution of the embedding vectors are compared between the existing learning method and the proposed learning method for verification. The verification was conducted using an open-set. The resulting accuracy of 97.33% for the proposed learning method is approximately 4% greater than that of the existing learning method.

Survei Penggunaan Tensorflow pada Machine Learning untuk Identifikasi Ikan Kawasan Lahan Basah

Wetlands are habitats commonly used for fish cultivation. South Kalimantan is one of the provinces that has a wetland area, which is 11,707,400ha, there are 67 rivers and an estimated 200 species of fish. This shows the abundant wealth of fish treasures and economic value. The study of fish identification is an important subject for the preservation of wetland fish. In the field of artificial intelligence, identification can be done using Machine Learning (ML). There are many libraries, a collection of functions that can be used in ML development, one of which is Tensorflow. In this paper, we survey a variety of literature on the use of Tensorflow, as well as datasets, algorithms, and methods that can be used in developing wetland area fish image identification applications.The results of the literature survey show that Tensorflow can be used for the development of fish character identification applications. There are many datasets that can be used such as MNIST, Oxford-I7, Oxford-102, LHI-Animal-Faces, Taiwan marine fish, KTH-Animal, NASNet, ResNet, and MobileNet. Classification methods that can be used to classify fish images include CNN, R-CNN, DCNN, Fast R-CNN, kNN, PNN, Faster R-CNN, SVM, LR, RF, PCA and KFA. Tensorflow provides many models that can be used for image classification, including Inception-v3 and MobileNets, and supports models such as CNN, RNN, RBM, and DBN. To speed up the classification process, image dimensions can be reduced using the MDS, LLE, Isomap, and SE algorithms.

The role of animal faces in the animate-inanimate distinction in the ventral temporal cortex

Animate and inanimate objects elicit distinct response patterns in the human ventral temporal cortex (VTC), but the exact features driving this distinction are still poorly understood. One prominent feature that distinguishes typical animals from inanimate objects and that could potentially explain the animate-inanimate distinction in the VTC is the presence of a face. In the current fMRI study, we investigated this possibility by creating a stimulus set that included animals with faces, faceless animals, and inanimate objects, carefully matched in order to minimize other visual differences. We used both searchlight-based and ROI-based representational similarity analysis (RSA) to test whether the presence of a face explains the animate-inanimate distinction in the VTC. The searchlight analysis revealed that when animals with faces were removed from the analysis, the animate-inanimate distinction almost disappeared. The ROI-based RSA revealed a similar pattern of results, but also showed that, even in the absence of faces, information about agency (a combination of animal’s ability to move and think) is present in parts of the VTC that are sensitive to animacy. Together, these analyses showed that animals with faces do elicit a stronger animate/inanimate response in the VTC, but that this effect is driven not by faces per se, or the visual features of faces, but by other factors that correlate with face presence, such as the capacity for self-movement and thought. In short, the VTC appears to treat the face as a proxy for agency, a ubiquitous feature of familiar animals.Significance StatementMany studies have shown that images of animals are processed differently from inanimate objects in the human brain, particularly in the ventral temporal cortex (VTC). However, what features drive this distinction remains unclear. One important feature that distinguishes many animals from inanimate objects is a face. Here, we used fMRI to test whether the animate/inanimate distinction is driven by the presence of faces. We found that the presence of faces did indeed boost activity related to animacy in the VTC. A more detailed analysis, however, revealed that it was the association between faces and other attributes such as the capacity for self-movement and thinking, not the faces per se, that was driving the activity we observed.

Object Categorization and Response Bias Flexibly Change Depending on Cross-Spatial Scale Coherence and Task Demands

Categorization of visual stimuli at different levels of abstraction (superordinate, basic, subordinate) relies on the encoding of relevant diagnostic features present at different spatial scales. We used the Eidolon Factory, an image-manipulation algorithm that introduces random disarray fields across spatial scales, to study how such a process flexibly combines perceptual information to perform successful categorization depending on task demands. Images of animal faces, human faces and everyday objects were disarrayed coherently (random fields correlated) or incoherently (random fields randomized) to create a family of 50 eidolons per stimulus image with increasing disarray. Participants (N=217) viewed each family of eidolons in a smooth sequence from maximum disarray to no disarray. Then, they performed a category verification task either at the superordinate (any face type) or basic (human face only) levels at different levels of uncertainty – in the first response, they used their gut feeling to respond, while they had to be sure of their decision in the second response. When participants used their gut feeling to respond, we observed a basic-level advantage, but not an effect of the type of disarray. When they were sure of their response, we observed a superordinate advantage and stronger disarray effects in the coherent stimulus. Furthermore, participants changed their decision criterion depending on the abstraction level of the task. These results suggest that the visual system flexibly adjusts to the relevant perceptual information depending on task context and that it does not strictly adhere to feedforward processing.

Yoginīs

Goddesses known as yoginīs (feminine of the Sanskrit yogin, “practitioner of yoga” or “possessed of yoga”) were prominent in the esoteric or “tantric” religious traditions of premodern India, beginning from around the 7th century ce. They feature especially in tantric Śaiva cults of Bhairava and allied goddesses, as well as within the Vajrayāna Buddhist Yoginītantras or Yoganiruttaratantras. In fact yoginī cults form a significant shared dimension of these tantric traditions. This bibliography is primarily concerned with yoginīs in premodern Hindu traditions, especially Tantric Śaivism, while nonetheless including some representative works on Buddhist and other traditions. While sharing characteristics with several other deity types, yoginīs have particularly deep connections with mātṛs or “mother-goddesses,” ancient deities associated with fertility, motherhood, disease, and warfare. Several key aspects of yoginīs are shared with the earlier mātṛs, such as the ability to fly, the high frequency of animal faces, occurrence in groups, martial prowess, and their simultaneous beauty and dangerous power. Yoginīs have particularly strong connections with the Gupta-era Seven Mothers (saptamātṛs or saptamātṛkās), who are often included in yoginī sets. Within Tantric Śaivism, yoginīs surface first in the Vidyāpīṭha division of Bhairavatantras, such as the extant Siddhayogeśvarīmata and Brahmayāmala, remaining prominent in Kaula traditions of the late 1st and early 2nd millennia. This literature depicts yoginīs as powerful, potentially dangerous flying goddesses who embody the numinous powers of yoga, powers sought by tantric practitioners through visionary, transactional encounters. Organized into clans (kula), yoginīs were regarded as both guardians and potential sources for the transmission of tantric revelation. While quintessentially tantric goddesses, the veneration of yoginīs took on more public forms by the 10th century. Temples dedicated to groups of yoginīs were constructed across India, mainly from the 10th to 12th centuries, and yoginīs also left their mark in non-tantric religious and narrative literatures. Although yoginī worship waned in the latermedieval period, these goddesses remained important in some tantric traditions, and have received renewed attention in the modern world. Distinctive to the figure of the yoginī is the blurring of boundaries between goddesses and women: in many contexts, the word yoginī simply refers to a female yogi or tantric initiate, and female adepts were viewed as potentially becoming divine yoginīs through sudden gnosis or ritual perfection. For this reason, while the bibliography is mainly devoted to scholarship on tantric goddesses, rather than yoginī in the more general sense of “female practitioner of yoga,” it also necessarily concerns female tantric adepts, gender and sexuality in the tantric traditions, and the impact of belief in yoginīs upon women. Indeed, one of many meanings of yoginī and closely related terms is “tantric sorceress” or “witch,” a notion carried into the modern world, sometimes with tragic consequences.

Becoming-Animal in the Narrative and the Form of Reha Erdem’s Kosmos

This article performs a narrative and aesthetic analysis of Reha Erdem’s movie, Kosmos (2009), through an engagement with Gilles Deleuze and Félix Guattari’s philosophical concept of becoming-animal. Erdem narrativizes the story of an odd traveller dervish named Kosmos, who has supernatural abilities and an expanded capability of communication—one that displays liminal features between human and animal. Through his distinctive editing technique, particularly by juxtaposing human and animal faces, the director further deconstructs the conceptual boundaries between humanity and animality, revealing the inherent connectedness of the two. Hence, this article discloses the consistency between the narrative and the form of Kosmos through a close reading based upon the notion of becoming-animal and its conceptual constituents.

Same father, same face: deep-learning reveals paternally-derived signalling of kinship in a wild primate

ABSTRACTAnimal faces convey important information such as individual health status1 or identity2,3. Human and nonhuman primates rely on highly heritable facial traits4,5 to recognize their kin6–8. However, whether these facial traits have evolved for this specific function of kin recognition remains unknown. We present the first unambiguous evidence that inter-individual facial similarity has been selected to signal kinship using a state-of-the-art artificial intelligence approach based on deep neural networks and long-term data on a natural population of nonhuman primates. The typical matrilineal society of mandrills, is characterized by an extreme male’s reproductive skew with one male generally siring the large majority of offspring born into the different matrilines each year9. Philopatric females are raised and live throughout their lives with familiar maternal half-sisters (MHS) but because of male’s reproductive monopolization, they also live with unfamiliar paternal half-sisters (PHS). Because kin selection predicts differentiated interactions with kin rather than nonkin10 and that PHS largely outnumber MHS in a mandrills’ social group, natural selection should favour mechanisms to recognize PHS. Here, we first show that PHS socially interact with each other as much as MHS do, both more than nonkin. Second, using artificial intelligence trained to recognize individual mandrills from a database of 16k portrait pictures, we demonstrate that facial similarity increases with genetic relatedness. However, PHS resemble more to each other than MHS do, despite both kin categories sharing similar degrees of genetic relatedness. We propose genomic imprinting as a plausible genetic mechanism to explain paternally-derived facial similarity among PHS selected to improve kin recognition. This study further highlights the potential of artificial intelligence to study evolutionary mechanisms driving variation between phenotypes.

Klasifikasi Wajah Hewan Mamalia Tampak Depan Menggunakan k-Nearest Neighbor Dengan Ekstraksi Fitur HOG

Mammal is a type of animal that has many diverse characteristics, such as vertebrates and breastfeeding. In this study, the HOG feature and the k-NN method were proposed to classify 15 species of mammals. This study uses the LHI-Animal-Faces dataset which has fifteen species of mammals, where each type of mammal has 50 images measuring 100x100 pixels. The image will be conducted the process by the HOG feature extraction process and continued into the classification process using k-Nearest Neighbor. The performance of the HOG and k-NN features that get the best value is in deer and monkey, the best results for precision, recall, and accuracy are at k=3 where HOG feature extraction provides good vector features to be used in the classification process using the k-NN method.

Klasifikasi Mamalia Berdasarkan Bentuk Wajah Dengan k-NN Menggunakan Fitur CAS dan HOG

Klasifikasi pada jenis objek sudah banyak dilakukan pada beberapa jenis data citra. Klasifikasi jenis hewan telah dilakukan menggunakan pendekatan segmentasi dan tanpa segmentasi sebagai tahapan awal. Context Aware Saliency (CAS) merupakan metode yang mampu membuat wilayah objek menjadi lebih dominan dibandingkan dengan background dalam mode saliency sehingga dapat menjadi alternatif pengganti proses segmentasi objek. Fitur bentuk diambil berdasarkan citra hasil saliency menggunakan metode Histogram of Oriented Gradient (HOG). Metode K-Nearest Neighbors (K-NN) digunakan untuk klasifikasi jenis hewan mamalia berdasarkan fitur HOG dari citra saliency. Dataset yang digunakan pada penelitian ini adalah LHI-Animal-Faces. Hasil yang didapatkan menunjukkan bahwa jenis hewan yang dapat dikenali dengan baik, yaitu Kucing dan Harimau, sedangkan Domba, Anjing, dan Babi belum mampu dikenali dengan baik.