Semi-Supervised Learning With Structured Knowledge For Body Hair Detection In Photoacoustic Image

Kikkawa, Ryo; Sekiguchi, Hiroyuki; Tsuge, Itaru; Saito, Susumu; Bise, Ryoma

doi:10.1109/isbi.2019.8759249

Cited by 12 publications

(9 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Sebelumnya ada beberapa penelitian yang telah dilakukan deteksi rambut, beberapa diantaranya adalah yang dilakukan oleh [8] dimana penelitian tentang deteksi rambut menggunakan metode machine learning, pada penelitian tersebut dapat disimpulkan jika metode yang digunakan dapat melakukan pemisahan antara rambut dengan kulit. Selain penelitian [8], ada juga penelitian tentang kondisi akar rambut di kepala dengan tanpa harus memotongnya terlebih dahulu, pada penelitian yang dilakukan [9] dapat diketahui jika kondisi dari akar rambut kepala dapat dideteksi kondisinya tanpa harus dilakukan pemotongan pada rambut kepala, adapun penelitian tentang rambut yaitu dilakukan oleh [10] dimana penelitian tersebut berawal dari tindakan medis yang memerlukan perhitungan rambut secara manual sehingga untuk memudahkan itu dibuat rancangan sistem untuk perhitungan rambut secara otomatis dengan dimulai dengan proses segmentasi rambut.…”

Section: Pendahuluanunclassified

Perbandingan Algoritma Multi-Thresholding, Konversi Biner, Low-Pass Filtering pada Segmentasi Rambut Kaki

Nurfalah¹,

Hadianti

Mayangky

et al. 2021

SISTEMASI

View full text Add to dashboard Cite

Rambut berada tersebar pada bagian tubuh manusia dan memiliki fungsi yang berbeda, untuk mementingkan estetika tidak sedikit orang yang melakukan pencukuran rambut pada beberapa bagian tubuh, salah satunya bagian kaki. Terdapat banyak metode pencukuran yang tersedia saat ini, mulai dari cara alami, terapi, modern sampai penggunaan laser, namun pada penelitian ini berfokus pada metode laser. Pencukuran rambut dengan metode laser kelebihannya dapat menghilangkan rambut dalam jangka waktu yang cukup lama dengan pengaplikasian yang cukup mudah, dan tidak memerlukan waktu yang lama dibandingkan dengan teknik pengaplikasian tradisional. Dengan permasalahan itu dibutuhkan sistem yang bisa membedakan wilayah rambut dan kulit, sehingga tujuan untuk hair removal lebih fokus pada bagian rambut dan tidak mencederai bagian atau lapisan kulit. Pemisahan dua bagian tersebut dilakukan dengan cara segmentasi helai rambut dan menghilangkan bagian kulit. Empat metode dibandingkan dalam penelitian ini yaitu multi-thresholding, Konversi biner, low-filter dan highfilter, hasil dari perbandingan empat metode diketahui jika metode Multi-Thresholding diyakini dapat melakukan segmentasi rambut dengan baik, karena pola rambut dapat terlihat jelas dan tidak banyak noise yang terlihat.

show abstract

Section: Pendahuluanunclassified

Perbandingan Algoritma Multi-Thresholding, Konversi Biner, Low-Pass Filtering pada Segmentasi Rambut Kaki

Nurfalah¹,

Hadianti

Mayangky

et al. 2021

SISTEMASI

View full text Add to dashboard Cite

show abstract

“…Thus, it is difficult to separate the regions by using a single threshold. To address this, we follow the detection method proposed by Kikkawa et al [3]. Because the intensity of the artifact is usually slightly less than the sound source (i.e., vessels and hairs), candidate regions are extracted by using multiple-level thresholds to separate all the regions.…”

Section: A Hair Candidate Region Detectionmentioning

confidence: 99%

“…In our observation, the position of hairs in the body is not correlated with the intensity feature of hairs; the regions of hairs placed near the surface have various intensity values, and the deeper area contains both lowand-high-intensity hairs. We simply use the intensity-based features proposed in [3]. We expect that the positive samples extracted in the previous step tend to be randomly distributed in this intensity-based feature.…”

Section: Feature Extraction For Pu Learningmentioning

confidence: 99%

See 1 more Smart Citation

Unsupervised Body Hair Detection by Positive-Unlabeled Learning in Photoacoustic Image

Kikkawa¹,

Kajita²,

Imanishi³

et al. 2021

2021 43rd Annual International Conference of the IEEE Engineering in Medicine &Amp; Biology Society (EMBC)

Self Cite

View full text Add to dashboard Cite

Photoacoustic (PA) imaging is a new imaging technology that can non-invasively visualize blood vessels and body hair in 3D. It is useful in cosmetic surgery for detecting body hair and computing metrics such as the number and thicknesses of hairs. Previous supervised body hair detection methods often do not work if the imaging conditions change from training data. We propose an unsupervised hair detection method. Hair samples were automatically extracted from unlabeled samples using prior knowledge about spatial structure. If hair (positive) samples and unlabeled samples are obtained, Positive Unlabeled (PU) learning becomes possible. PU methods can learn a binary classifier from positive samples and unlabeled samples. The advantage of the proposed method is that it can estimate an appropriate decision boundary in accordance with the distribution of the test data. Experimental results using real PA data demonstrate that the proposed approach effectively detects body hairs.

show abstract

“…Semi-supervised learning (SSL): SSL is one of the most promising approaches to improve the performance of classifiers using unlabeled data. Most SSLs take a pseudo label learning approach [6,19,16,30]. For example, pseudo labeling [19] (also called self-training [36]) first trains using labeled data and then confident unlabeled data, which is larger than a predefined threshold, is added to the training data and the classifier is trained iteratively.…”

Section: Related Workmentioning

confidence: 99%

Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology

Tokunaga

Iwana

Teramoto

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

In pathological diagnosis, since the proportion of the adenocarcinoma subtypes is related to the recurrence rate and the survival time after surgery, the proportion of cancer subtypes for pathological images has been recorded as diagnostic information in some hospitals. In this paper, we propose a subtype segmentation method that uses such proportional labels as weakly supervised labels. If the estimated class rate is higher than that of the annotated class rate, we generate negative pseudo labels, which indicate, "input image does not belong to this negative label," in addition to standard pseudo labels. It can force out the low confidence samples and mitigate the problem of positive pseudo label learning which cannot label low confident unlabeled samples. Our method outperformed the state-of-the-art semi-supervised learning (SSL) methods.

show abstract

Semi-Supervised Learning With Structured Knowledge For Body Hair Detection In Photoacoustic Image

Cited by 12 publications

References 4 publications

Perbandingan Algoritma Multi-Thresholding, Konversi Biner, Low-Pass Filtering pada Segmentasi Rambut Kaki

Perbandingan Algoritma Multi-Thresholding, Konversi Biner, Low-Pass Filtering pada Segmentasi Rambut Kaki

Unsupervised Body Hair Detection by Positive-Unlabeled Learning in Photoacoustic Image

Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology

Contact Info

Product

Resources

About