Face recognition transfiguration film artifact, help house male pictures to search pornography anchor
Recently, a porn broadcast company launched a new otaku must-have feature: adult version to search for anchor. The user may select an erotic anchor similar to the one in the photo by uploading a photo.
Megacams’ promotional copy is as follows:
Everyone has a special person in their hearts. They hope to find a similar person on the platform to satisfy their fantasies. However, text search is not convenient. Fortunately, we have image technology. Face recognition is the best solution for him/her.
According to official sources, there are 180,000 people in Megacams' anchor library, and an average of 5,000 people are online at the same time. Users only need to upload a portrait photo, and Megacams will find similar anchors based on face recognition technology and give similarity.
For user privacy protection, Megacams does not store any user-uploaded photos. After searching out the results, the photos are automatically deleted.
Of course, there is a serious privacy issue here. If you try to take a picture of your friend's search, if you really find out, she wouldn't be able to. At the same time as it is convenient for users, it is difficult to ensure the privacy of the anchor.
Curious users take this opportunity to try, but find that the current match is not high. Among them are users who uploaded their own photos:
The search results are as follows:
The highest similarity anchors were 46%, 42% and 40% respectively.
There are several reasons why the match is not high:
The first is the quality of the image uploaded by the user: if it is not HD, pixel blurring will cause the recognition result to be biased.
The second is the limited number of anchors, making the matching library not exceed 20,000 and the base number is low.
Finally, the user uploads too few photos: If the user uploads more photos of a person (such as a person's 100 high-definition, 360-degree without dead ends, photos of various expressions), the more favorable the Megacams to the person's face Features are extracted so that Megacams more fully and comprehensively grasps the character appearance information to more accurately match the anchor.
Actually, the algorithm of Megacams is actually similar to Google’s and Baidu’s “Mapping†function, which is roughly as follows:
Face Detection: Find the location of the face in the image.
Face feature point location: including the eyeball center point, cheeks and other outline feature points.
Human face geometry normalization: The normalized face area image is obtained from the image according to the position of the facial feature point. This step is mainly to make the face position corresponding to the pixels of different faces uniform.
Face feature extraction, mapping high-dimensional face features to low-dimensional features with stronger classification capabilities.
The distance between the discriminative features obtained by the two faces is calculated and used as a measure of similarity.
Although the face recognition algorithm has been greatly improved in recent years, the degree of intelligence is still lacking. At present, search by graphs is excellent when looking for the same person, but there are still some problems in searching for similar people. The word "similarity" itself has the subjective color of human beings. Therefore, the similarity is obtained only through feature extraction and keypoint positioning, and it is still worse.
Related Reading:
Face Recognition New Application Scenario: Preventing a Minor from Entering Internet Cafes
The Application of Deep Learning in Face Recognition——The Evolution of Grandma Grandmother Model
Washington’s latest research: How Face Recognition Algorithms Can Do When Faced with 1 Million Unfamiliar Facebook Faces
Coaxial cable can be divided into baseband coaxial cable and broadband coaxial cable (namely network coaxial cable and video coaxial cable). Coaxial cables are classified into 50 ω baseband cables and 75 ω broadband cables. Baseband cable is divided into thin coaxial cable and thick coaxial cable. Baseband cables are used only for digital transmission, with data rates up to 10Mbps.
Coaxial Cable means a Cable with two concentric conductors sharing the same axis as the shield. The most common type of coaxial cable consists of a copper conductor separated by an insulating material. Outside the inner insulation is another ring of conductors and their insulators. The entire cable is then wrapped in a PVC or teflon sheath.
Coaxial Cable,Rf Coaxial Cable,Coaxial Cable Connectors,Tv Coaxial Cable
CHANGZHOU LESEN ELECTRONICS TECHNOLOGY CO.,LTD , https://www.china-lesencable.com