Method for determining a perfume according to physiognomic features of a person's face

Abstract

A method for determining a perfume according to physiognomic features of a person's face consists in uploading to a database information about perfumes and their constituent aroma compounds, each of which corresponds to a top, middle or base note, and information about the correlation between facial features and aroma compounds, uploading a picture of the face of a person for whom a perfume is to be selected, identifying the characteristic physiognomic features of the person's face, which are then used to find people with similar features in the database, selecting aroma compounds that correspond to the characteristic physiognomic features of the face of the person on the uploaded picture, and selecting for said person a perfume containing the chosen aroma compounds, fragrance compounds and essences, each of which corresponds to a top, middle or base note, on the basis of an image from the database which most closely matches the face of the user. The technical result consists in more accurately selecting a perfume for the user.

Description

[0001] METHOD FOR DETERMINING PERFUME BY PHYSIOGNOMIC CHARACTERISTICS OF A HUMAN FACE The invention relates to the field of recognizing the physiognomic characteristics of a human face and determining the aromatic, fragrant and essential substances that make up a perfume aroma, with the aim of identifying the most suitable perfume for a specific person, taking into account the identified physiognomic features.

[0002] Choosing a fragrance that complements and reflects a person's personality is a challenging task, most often accomplished in perfume stores with the help of a consultant who learns the customer's preferences and, based on their knowledge of the store's products, suggests fragrance options that match the customer's preferences. In this case, the successful selection of a fragrance that suits the customer depends on the experience and knowledge of the sales consultant, making the decision highly subjective.

[0003] The current goal is to simplify and automate the process of selecting a perfume that suits a specific person, taking into account their clothing style, season, gender, age, hair length, eye color, and other characteristics. This process also involves selecting and taking into account the odorant notes contained in each perfume.

[0004] Perfume note is a class in the classification of essential oils and other aromatic, fragrant and ethereal substances (hereinafter referred to as aromatic substance) based on their rate of evaporation.

[0005] Perfume notes are divided into the following types: top (head, initial), middle (main, heart notes) and base (background, final).

[0006] Top notes are aromatic substances that evaporate quickly within 30 minutes (for example, lemon essential oil), middle or “heart” notes are those that evaporate moderately within 3-6 hours (for example, rose essence), and base notes are those that evaporate slowly within 24 hours (for example, sandalwood).

[0007] Perfume (or perfume composition, or perfume fragrance) is any mixture of aromatic substances or a combination of perfume notes. It is the result of the work of a perfumer who has created a complex blend of aromatic substances or perfume notes.

[0008] A perfume composition is also often referred to as a "chord," which in turn consists of three sets of "notes."

[0009] In other words, perfume notes are arranged in perfume by type of perfume note (base note, middle note, top note), thus determining the durability of the scent or the duration of the scent (Internet resource htps: / / aroma-intuition.ru / articles / stoykost-aromatov#:~:text=%D0%98%D0%B7%20%D0%BF%D0%B0%Dl%80%Dl%84 %Dl%8E%D0%BC%D0%B5%Dl%80%D0%BD%D0%BE%D0%B3%D0%BE %20%D 1 %83%D 1 %87%D0%B5%D0%B 1 %D0%BD%D0%B8%D0%BA%D0% B0%20%D0%9A%D 1 %81 %D0%B5%D0%BD%D0%B8%D0%B8%20%D0%90 %D0%BB%Dl%8C%Dl%82%D0%B5%Dl%80,%2C%20%Dl%82%D0%BA% D0%B0%D0%BD%D0%B8%20%D0%B8%20%D0%B4%Dl%80, published 02 / 21 / 2024, viewed 10 / 16 / 2024).

[0010] An important factor is that the same aromatic substance can appear in different types of notes in different perfumes, giving each product its unique character. For example, sandalwood can be both a middle note and a base note (https: / / aromo.ru / notes / wood / sandal / , published on July 2, 2021, viewed on October 16, 2024), while orange can be present in both the top note and the middle note (https: / / randewoo.ru / statiy / citrusovye-noty-v-parfiimerii.html, published on April 28, 2013, viewed on October 16, 2024). That is, when choosing a perfume for a specific person, it is important to determine not only the list of aromatic substances included in it, but also specifically what type of perfume note (top, middle, base) a particular aromatic substance belongs to.

[0011] A prior art method for identifying perfumes based on the characteristics of a specific user is known (online resource https: / / aromo.ru / articles / essays / perfume-apps-review / , published September 22, 2022, viewed October 16, 2024). This method involves uploading user information to a database, including gender and age, preferred fragrances, personality type, fashion and design preferences, lifestyle, season and price range, purpose of using the perfume (office, party), and style (classic, eclectic). Information about the notes present in each perfume, a description of the perfume, and customer reviews are also uploaded to the database. The verbal information about the user is then compared with verbal information about perfume products stored in the database, the perfume with the closest match is identified, and suitable perfume options are determined.

[0012] This type of perfume selection is implemented using the Sommelier du Parfum app. This method automates and simplifies the process of choosing the right perfume. However, a drawback of this method is that accurate results depend on the amount of information entered about the perfume's characteristics. If customer reviews are insufficient (for example, the product has just hit store shelves) or the description of the perfume's notes is incomplete (for example, the manufacturer has not provided complete information), then a verbal comparison of the user's preferences and the perfume information may result in few matches, preventing the selection of the desired perfume from the store's products. In this case, the accuracy and effectiveness of matching the perfume to the user's request will be extremely low.In addition, the comparison of information occurs only on the basis of verbal information about the user, which may differ from reality, especially in matters of appearance characteristics, which again will affect the inaccuracy of the obtained result.

[0013] A prior art method for identifying perfumes based on facial characteristics (Korean Patent No. 102495868, published February 6, 2023) is considered the closest analogue to the claimed solution. This method involves storing multiple fragrance series in a database, along with information about the trend, product type, product characteristics, fragrance series type, and fragrance series intensity. Next, a camera is used to upload an image of the face of the person for whom a perfume is to be selected. Review information is displayed for each perfume listed on the customized perfume list page based on a predefined facial expression object, which includes expression objects such as expressionless, unstable, suspicious, confused, frightened, and urgent. A personalized perfume list page is created based on the product information and whether the user's fragrance series is the same.Using a neural network, they determine the shape of the eyelids in a facial image and recommend anti-drowsiness perfumes by integrating them into a customized perfume list page.

[0014] This method allows you to determine the list of aromatic substances that "suit" specific facial features of a person, and, taking this into account, select a perfume that contains exactly the same aromatic substances.

[0015] However, this method doesn't take into account the specific note type (top, middle, base) of the fragrance. This means the user may be offered perfumes where, for example, orange oil is found in the top or middle note. Therefore, this method doesn't accurately determine the specific note type of each selected fragrance, making perfume selection less accurate.

[0016] The technical result of the invention is the creation of a method for determining perfume based on the physiognomic characteristics of a person's face, allowing for a more accurate selection of perfume for the user.

[0017] The technical result is achieved by using a method for determining perfume based on the physiognomic characteristics of a person’s face, which consists in the following:

[0018] - information about perfumes and the aromatic, fragrant and essential substances that comprise them is loaded into the database, each of which corresponds to a top, middle or bottom perfume note;

[0019] - information about the correspondence between the characteristics of people's faces and aromatic, fragrant and essential substances, each of which corresponds to a top, middle or bottom perfume note, is loaded into the database;

[0020] - upload an image of the face of the person for whom you need to select a perfume;

[0021] using a neural network, the features of the physiognomic characteristics of a person's face are determined based on the uploaded image;

[0022] - using software, a comparison is made between the physiognomic characteristics of a person's face from a downloaded image and the characteristics of people's faces from a database, and people with similar physiognomic characteristics of a face are found in it;

[0023] - using software, aromatic, fragrant and essential substances are selected from the database, each of which corresponds to the top, middle, or bottom perfume note, and which at the same time correspond to the features of the physiognomic characteristics of the person’s face in the uploaded image;

[0024] - using software from a database, a perfume is selected that contains selected aromatic, fragrant, and essential substances, each of which corresponds to a top, middle, or bottom perfume note.

[0025] Information about a perfume includes its name and manufacturer. It may also include the price, location on store shelves, and other relevant information.

[0026] The information about the perfume notes in each perfume includes information about the presence of specific aromatic substances in it, each of which corresponds to a top, middle, or bottom perfume note.

[0027] A human face image is a photograph, multiple photographs, a still image from a video, a video image, or similar images that display a human face.

[0028] An image of a person's face is uploaded to the server from any device or gadget: telephone, computer, laptop, smartphone, etc. A database containing information on the correspondence between the characteristics of a person's face and aromatic substances is formed in advance using information from open sources, the Internet, etc. The database can also be compiled through the joint work of many specialists (physiognomists, perfumers, and other specialists in the field of cosmetology, perfumery, medicine, etc.), who obtain objective data based on scientific research (for example, Internet resources htps: / / www.kommersant.ru / doc / 6117957, htps : / / www.sbras.info / public-reception / question / fiziognomika-lzhenauka).

[0029] A similar database generation method, as used in Korean Patent No. 102495868, which is considered the closest solution to the claimed method, can also be used. The neural network determines physiognomic characteristics of a person's face from an image, including absolute and relative facial parameters (the length of facial zone segments defining the height, width, and depth of facial zones, as well as the angles of facial zones), the presence of wrinkles, the presence of scars, and eye color.

[0030] The neural network also determines skin color, the presence of piercings, tattoos, makeup, emotions, hair length and color, age, gender, race, and more.

[0031] In addition, when choosing a perfume, the price range, time of year, clothing style, and other preferences of the user can be taken into account.

[0032] The availability of a database of available perfumes and a database of facial characteristics allows for the generation of the necessary and sufficient information for further objective and automated analysis of the user's facial features using a neural network and software. This allows for a more accurate determination of the appropriate aromatic substances for the user and their specific note type (base, middle, top). This ultimately allows for a more personalized selection of perfumes that take into account all the user's preferences and wishes, as well as exquisitely highlighting and complementing their appearance.

[0033] A neural network trained on facial images and publicly available internet data allows it to quickly and accurately read and identify a user's facial features. Each use of the neural network also continues its learning process, adding new data to its database.

[0034] Identifying the aromatic substances that are suitable for a person and determining their type of note (top, middle, base) allows for a step-by-step, more thorough and precise approach to emphasizing the individual features of the physiognomic characteristics of a person's face, including taking into account age.

[0035] In general, the use of all the processes involved in the claimed method helps to increase the speed of data processing, eliminate the subjective factor, and increase the accuracy of determining the result - the selection of a suitable perfume, where aromatic substances are arranged taking into account their correspondence to the type of notes (base, middle, top).

[0036] Fig. 1 shows an image of the gadget screen, where the user 10 specifies the parameters for selecting a perfume.

[0037] Fig. 2 shows an image of the user's face with the identification of physiognomic characteristics of the face in the uploaded photograph using a neural network.

[0038] Fig. 3 shows an image showing the selection of an image from the database that best matches the user's face in the uploaded photograph.

[0039] Fig. 4 shows a table of suitable aromatic substances, each of which corresponds to a base, middle, or top perfume note.

[0040] 20 Fig. 5 shows an image of the gadget screen with a perfume suitable for the user.

[0041] The claimed method is implemented as follows.

[0042] The database is loaded with information about perfume fragrances, as well as information about the aromatic substances contained in each perfume, taking into account each aromatic substance's correspondence to the base, middle, or top note. Information about the perfume's description, including characteristics by season, clothing style, and price range, is also uploaded.

[0043] A database of facial parameters (images) is also being created. Each image (face) in the database is associated with aromatic substances, indicating their correspondence to the type of perfume note (base, middle, top) that suits the given appearance.

[0044] This database of aromatic substances, taking into account their type of notes (base, middle, top) and appearance characteristics, is compiled in advance with the participation of a perfumer, physiognomist, and other specialists, as well as using open-access sources.

[0045] Next, the user uses the app to select their preferred clothing style, price range for the perfume, and the season for which they want to select the perfume.

[0046] Initially, perfume fragrances from the database are filtered according to these parameters by comparing the user's wishes and the information in the perfume cards.

[0047] Then, an image of the user's face is uploaded to the database, which is taken using the camera of any gadget or downloaded from the gadget's gallery of saved photos.

[0048] Using software and a trained neural network, visual facial characteristics are analyzed from an uploaded image and physiognomic data is converted into numbers.

[0049] Using software, the numerical data of the user's facial characteristics, determined by the neural network, is compared with a database created in the service's admin panel or created in any other way.

[0050] Using software, the database is used to find the most similar image (face) to the image (face) from the uploaded image.

[0051] Next, the fragrances associated with the user's face image in the admin panel are compared with the fragrances of the perfumes loaded into the corresponding database. A perfume is then found that contains not only the necessary fragrances suitable for the specific user, but also the specific arrangement of each of these fragrances, corresponding to the type of perfume note (base, middle, top).

[0052] Ultimately, the gadget's screen displays a result to the user in the form of information about the perfume that most closely matches the parameters of the uploaded person's face, or several results with an indication of the percentage match.

[0053] Users can view a full description of a perfume or make a purchase in the INDIP app, the INDIP web version, or through a partner service with which the system interacts, or they can visit the manufacturer's or retailer's website.

[0054] Using the claimed method, if the user prefers orange oil, located in the middle note, and sandalwood oil, also located in the middle note, they will be matched with a perfume where these aromatic substances are located in the middle note. This result of implementing the claimed method demonstrates that it is more accurate than the method based on the closest analogue. In the standard version, the user would be matched with a perfume where orange oil is located in the top note and sandalwood oil in the base note, which would be inaccurate.

[0055] Example 1

[0056] The user in the INDIP application, when using any gadget, selects the parameters: clothing style (casual, business, sports, classic), season of the year and price range (Mass, Middle, Premium, Lux) (Fig. 1).

[0057] The user (girl), according to the example, chooses a premium perfume for everyday use in the autumn period.

[0058] She uploads a photo of herself from her device's gallery or takes a picture of her face in the INDIP app. Then, using a neural network, the physiognomic characteristics of her face are determined, specifically, the facial points, angles, and proportions, the presence of wrinkles, scars, etc. (Fig. 2).

[0059] Then, using software in the admin panel, they find a face (image) that is as similar as possible in percentage terms to the user’s face based on the location of points, angles, distances, and other values ​​determined by the neural network (Fig. 3).

[0060] In the example under consideration, the degree of match with the image from the database is 89%.

[0061] Specific aromatic substances are linked to the face found in the database that best matches the user's facial parameters in the admin panel, taking into account the belonging of each such substance to a specific type of note (Fig. 4).

[0062] Taking into account the information received from the database, the result is determined in the form of one main and several additional perfumes that are most suitable for the user in order of the highest percentage match (Fig. 5).

[0063] The result is displayed on the gadget screen.

[0064] In the example under consideration, the main perfume fragrance chosen for the girl was YSL Libre, and the alternative options were KENZO, LE LABO, VERSACE, and OJAR.

[0065] f Thus, the claimed method allows for the precise and quick selection of the required combination of aromatic substances, with each substance precisely matching the type of perfume note, which, during the process of using the perfume, "reveals" the perfume aroma, emphasizing the individual characteristics of the user's image.

Claims

CLAUSES OF THE INVENTION A method for determining perfume based on the physiognomic characteristics of a person's face, which consists of: - information about perfumes and the aromatic, fragrant and essential substances that comprise them is loaded into the database, each of which corresponds to a top, middle or bottom perfume note; - information about the correspondence between the characteristics of people's faces and aromatic, fragrant and essential substances, each of which corresponds to a top, middle or bottom perfume note, is loaded into the database; - upload an image of the face of the person for whom you need to select a perfume; using a neural network, the features of the physiognomic characteristics of a person's face are determined based on the uploaded image; - using software, a comparison is made between the physiognomic characteristics of a person's face from a downloaded image and the characteristics of people's faces from a database, and people with similar physiognomic characteristics of a face are found in it; - using software, aromatic, fragrant and essential substances are selected from the database, each of which corresponds to the top, middle, or bottom perfume note, and which at the same time correspond to the physiognomic characteristics of the person’s face in the uploaded image; - using software from a database, a perfume is selected that contains selected aromatic, fragrant, and essential substances, each of which corresponds to a top, middle, or bottom perfume note.

Images