Supports 72 key points, 150 key points, and 201 key points of face detection. Key points include face, eyes, eyebrows, lips and nose contour, etc. This service has the following three business functions:

Adopting advanced adversarial generative network technology, it breaks through the "next-dimensional wall" with one click, retains user characteristics in multiple dimensions, and achieves the effect of a thousand faces; with a variety of comic style image migration, it generates highly cute comic faces with artistic beauty for users. Including Pixar, 3D cartoon style, Angel, Demon, Ukiyo-e, American comic style, Pop style, Japanese manga style, Chinese cartoon style, Retro comic style, Cute cartoon style, Hand-drawn style, Pencil drawing style, Artistic effects, etc.

Based on deep learning algorithm, it can add bangs, change long hair, increase the number of hair and other operations to the hairstyle of portrait. In addition to helping users intuitively experience a variety of hair designs and improving the personalized experience of customers in the beauty and hairdressing industry, it can also be used in short videos, social platforms, or integrated into album-type apps to add hair editing play to users' personalized photos to achieve interactive participation and sharing by users.

Through AI technology, it detects the skin color of the face in the picture, supporting 1-6 stops, the smaller the number the lighter the color.

Edit the facial attributes, realize special effects such as gender swap and age change, and generate a variety of special effects photos for users.

The virtual makeup change service breaks through the limitations of the custom fixed bottom plate color selection in the existing makeup testing technology. It can transform tens of thousands of target makeup pictures, realize thousands of people and greatly reduce the introduction, development and maintenance costs. It has unique advantages in live video delivery, skin care & beauty teaching scenarios, and can be widely used in offline beauty stores, online smart medical beauty, interactive entertainment, network marketing and other scenarios.

Provides multi-dimensional detailed skin analysis on the skin, comprehensive detection and identification of skin color, skin smoothness, acne spots, wrinkles, pores, blackheads, dark circles and eye bags, etc., accurate to such as blood vessel dark circles, acne marks, etc. detail

According to the specified color palette or automatic color extension of the picture, avoid abnormal color extension in visual hot spots.

Dehaze the images with unrecognizable details when shooting in dense fog to restore clearer and truer images.

Adjust the contrast of an image that is too dark or too bright to make the image more vivid.

Automatically identify over-stretched images and restore the image content to normal proportions.

Realize the intelligent and fast denoising of the compressed blurred image, optimize the image texture details, and make the picture more natural and clear.

Intelligently recognize the content of black and white images and fill in colors to make black and white images come alive.

The image can be converted into cartoon painting, pencil painting, colored pencil painting, or Gothic oil painting, colorful candy block oil painting, shouting oil painting, Kanagawa surfing oil painting, strange oil painting, lavender oil painting, etc. 9 styles, which can be used to carry out fun activities or integrate Change the style of the image in the Meitu app.

Using the world's leading confrontation generation network, combined with face detection, hair segmentation, portrait segmentation and other technologies, tailor-made two-dimensional animation images with thousands of people and thousands of faces for users, and can generate two-dimensional animation portraits wearing masks through parameter settings.

The two faces are fused, and the generated face has the appearance characteristics of the two faces at the same time (not changing the face).

The sky area and the hair area of the characters included in the input image can be processed by static rotation, and the micro-moving AVI format video can be obtained.

The color of the reference picture can be transferred to the original image, and the color transfer of high-definition images can be supported, which can ensure that the color of the portrait part does not change.

Algorithmically toning the input image, and outputting the toned image.

The image cropping capability can transform the input image according to the specified size. Support to automatically determine the position of the subject area, and use a better cropping method to crop the image.

The intelligent composition capability can intelligently output bounding boxes through aesthetic evaluation of the input image, and according to these bounding boxes, the original image can be cropped into a better image.

The photo editing capability can transfer the style of the input reference image, such as lighting and color, to the target image without affecting the structure of the original image.

Image invisible image watermarking capability can add or parse image watermarks to images.

The image invisible text watermark capability can add or parse the specified text watermark to the image.

The Image Human Erase capability can erase portraits in specified areas of an image and automatically fill in the background.

The image composition aesthetic scoring ability can score the composition aesthetics of the input image. The higher the score, the better the composition effect.

The image exposure scoring ability can score the exposure of the input image. The higher the score, the greater the exposure.

The style transfer capability can transform the style of the input image, so that the visual style of the image such as color and brush strokes can be transformed.

Automatically recognises the outline of the visually centred subject in the input image, uses the subject as the foreground and then erases the background; returns the foreground subject image (4 channels) after erasing the background, suitable for background removal scenes such as people, animals, food and objects.

Recognises the human silhouette in the input image, removes the background and returns the foreground portrait image (4 channels). Suitable for scenes with single/multiple people, complex backgrounds, various types of human poses, etc.

Automatically identifies the product outline in the input image, separates it from the background and returns a foreground product image (4 channels) with the background removed, suitable for single/multiple product, complex background scenarios, etc.

Identifies food areas in the input image and performs pixel-level background removal of food items in the image.

Identifies the outline of the car in the input image, performs pixel-level background removal of the car and outputs a transparent image.

Enter a furniture related image and perform a pixel level background removal of the furniture in the image.

Recognises the outline of the human head in the input image, including the face, hair and ears, and hairpiece area, without the neck, returning a transparent image (4 channels) with only the head area visible, suitable for single person scenes, and multiple person scenes. The input of images with a more visible portrait will work better.

AI-based algorithms automatically perform skin leveling and blemish removal (acne, pimple marks, freckles, etc.) on face areas, as well as skin area whitening on the whole body, while preserving skin texture as much as possible. Supports multi-person image processing.

Based on AI algorithm to automatically identify the person area in the image and generate sketch painting effect.

Based on AI algorithm to optimize the beauty of the face in the image, it supports a variety of effects such as peeling, removing dark circles, lines, and whitening.

Based on AI algorithm to simulate realistic makeup effect, it further enhances the face beautification effect by adding lipstick, highlighter, whole makeup and other makeup materials. Users can choose different beauty types and match with personalized beauty materials to complete the makeup look.

Based on AI algorithm to intelligently identify the range of faces in the image and blur the faces in the image.

Based on AI algorithm to adjust the five facial parts in detail, the adjustment content is: face, eyes, nose, mouth, other, etc.

Based on computer vision technologies such as 3D face and adversarial generative networks, the image fusion operation is performed on the portraits in the stencil image and the fusion image, and the fused image is returned.

Using advanced image processing technology, the input face picture is intelligently changed into beauty. It can help the beauty ability of cell phone manufacturers, beauty apps and other camera classes to achieve sharpening of skin, skin tone whitening, face slimming, adjustment of five features, spot and acne treatment in one click intelligently. Can also be applied to interactive entertainment and other scenarios, such as live streaming, short video, social networking platforms, easily enhance the user image.

Based on AI algorithm to transform the human eyes in the image, i.e. closed eyes to open eyes, the effect is realistic and natural. The main application scenarios are: intelligent photo restoration (through the eyes closed to eyes open technology, restoring the closed-eye photos captured to restore the moment of celebration), interactive entertainment (special effects production, to attract users to interact, participate and share).

Use advanced image processing technology to edit the expressions of human faces in images. Supports application scenarios such as real-time effects and photo editing.

Based on face driving algorithm, it can turn the face in the image into dynamic effect. It can drive the five senses in the image to do any action, make the face vivid, turn static into dynamic, and make the user's photo full of dynamic feeling. The difference with expression editing is that this ability supports any face to generate all kinds of dynamic and complex movements, and can make long time dynamic expressions. It can be used for online entertainment video generation, expression pack production, etc.

Based on portrait segmentation technology, it realizes 2D to 3D conversion of photos to achieve the effect of 3D naked eye slow motion. The selfie image and the background image using Hitchcock technology senseless fusion, open the scene stretching video shooting new way to play, can be used for personal social entertainment, film and television drama effect simulation shooting, etc.

The input natural skin images are used to predict the classification of skin diseases, and common diseases such as `acne`, `actinic keratosis`, `alopecia androgenetica`,` alopecia areata`, `bullous dermatosis`, `chloasma`, `corn`, `dermatofibroma`, `eczema dermatitis`, `erysipelas`, `erythema multiforme`, `folliculitis`, `furuncle`, `haemangioma`, `herpes`, `herpes` `simplex`, `iga vasculitis`, `keloid`, `keratosis follicularism`, `lichen planus`, `lupus erythematosus`, `molluscum contagiosum`, `nevus`, `paronychia`, `pityriasis alba`, `pityriasis rosea`, `prurigo nodularis`, `psoriasis`, `rosacea`, `sebaceous cyst`, `sebaceousnevus`, `seborrheic dermatitis`, `seborrheic keratosis`, `skin tag`, `stasis dermatitis`, `cyringoma`, `tinea capitis`, `tinea corporis`, `tinea cruris`, `tinea manuum`, `tinea pedis`, `tinea unguium`, `tinea versicolor`, `urticaria`, `urticaria papular`, `varicella`, `verruca plana`, `verruca vulgaris` and `vitiligo` can be detected.

Provides detailed multi-dimensional skin analysis of skin, comprehensive analysis of skin condition, such as skin color, skin texture, double eyelids, eye bags, dark circles, wrinkles, acne, spots, etc.

Automatically identifies the outline of the animal in the input image, separates it from the background and returns a foreground image of the animal with the background removed.

Identifies areas of clothing in the input image and performs pixel-level background removal of clothing in the image.