国产一区二区激情,精品国产乱码一区二区三区网址,精品一区二区三区四区五六区

久久久精品一区中文字幕,国产一区欧美一区二区日韩,国产一区二区成人久久919色,曰本一区二区三区精品免费,国产日产久久高清欧美一区,操天天操天天干天天,国产99久久久产精品免费,精品久久久中国字幕,99精品久久久久免费看蜜臀

Consumer Electronics

Consumer Electronics Application Development

Visual Perception and Image Processing

Development and Applications

Utilizing traditional image processing algorithms, AI image processing algorithms and biometrics recognition algorithms, providing our clients on smart phones and mobile terminals an integrated solution based on image processing and computer vision, actively participating in an open community and empowering the industry towards a smarter future.

High-quality One-stop Services

for Customers

The business department has more than 100 employees. Relying on closed-loop services including algorithms development, integration, debugging, optimization, testing and image quality delivery teams, it can provide customers with high-quality one-stop services. At present, more than 20 image processing algorithms have been implemented and commercialized in some leading mobile phone brands. Representative commercial terminals include Samsung, Lenovo, Hisense, MOTO, Sharp, Kyocera, Black Shark and many overseas operators.

Dual-camera Bokeh

To achieve a better bokeh effect via
computing the disparity of two cameras
for the same objective in the 3D space.

· Support RGB+RGB. RGB+MONO.Main+Wide

· Support real-time preview and photo taking

· Support adjustment of bokeh strength

· Natural light spot effect, which brightness, density and shapes can be adjusted

· Support refocusing and adjustment of bokeh strength in album

Single Camera Portrait Bokeh

Via massive data from selfie, extracting
the contours and blurring the background
then composites them together to achieve
portrait bokeh effect in a single camera.

· Support real-time preview and photo taking

· Support adjustment of bokeh strength

· Precise segmentation, especially deals with difficult parts like fingers, portrait edge

· Support portrait of a crowd background blurring

Portrait Video Bokeh

Based on neural network and massive
data training, achieve accurate extraction
of portrait in every frame and the background
is segmented for bokeh rendering of consecutive
frames, and finally realize the portrait video bokeh
effect.

· Support real-time preview and video recording.

· Support adjustment of bokeh strength

· Support multiple video input and video record formats

Light Beautification

Identifying the skin tone areas in photos, increase skin brightness and adjust skin color, then perform smoothing skin and freckle removal to achieve natural beautification of the characters.
Based on the recognition of 108 key facial features, combined with the standards of face aesthetics, change shape of face by adjusting these key facial features.

· Skin brightness adjustment

· Skin color adjustment including whitening, rosy and tanning

· Smooth texture of the skin finely while protecting the facial features

· Skin freckle removal adjustment

· Binocular size adjustment

· Adjustment of fat and thin cheeks

· Support high resolution video shooting

Face Unlock

Based on deep learning on massive facial data,
achieve face identification and liveness detection,
then the face features recorded by the device are
compared to input face features to realize the face
unlocking.

· Advanced liveness detection technology, resist photo, screen, 3D dummy attack

· Liveness detection while entering

· Support eye closure detection

· Support face unlock of various skin colors and races

· Support fast face unlocking in various angles and extreme environments

High Dynamic Range Imaging

HDR (High-Dynamic Range )， HDR (High-Dynamic Range) image, which means the intensity of light varies greatly.
By capturing multiple frames of images with different exposure, the final HDR image is composited by using the best details corresponding to each frame image to provide more dynamic range and image details.

· Real-time detection of high dynamic range scenes

· Provide appropriate exposure value according to detection status to achieve dynamic exposure

· Suppression of highlights, brightening the shadows and restore details of captured images

· Anti-ghosting for moving objects to achieve capturing clear images

· The skin color of the portrait is truly restored (protected)

· Use multiple frames to reduce noise

· Dynamic EV exposure

Multi-frame Noise Reduction

By continuously taking multiple frames of YUV/RAW photos, selecting the best frame, via comparing and compositing multiple frames to realize a noise reduction effect with extremely low image quality loss.
Compared with platform algorithms, the noise
reduction effect is better and easier to maintain.

· Reduce noise and improve resolution

· Improve color reproduction

· Anti-ghosting, anti-shake

· Increase brightness of dark area but never be over-exposure

· Support YUV and RAW data input

Super Night View

Bursting multiple frames of YUV/RAW
images, via compositing multiple frames
to achieve brightening and noise reduction,
and then processing with HDR algorithm,
to obtain clearer hand-held night scene photos.

· Improved image detail, low noise, high color reproduction

· Suppress highlights and brighten shadows separately to increase image detail

· Anti-ghosting, anti-shake, support handheld capturing

Super Resolution

Capture 3 to 8 frames of images for
compositing, and then output photos
with better details, which can improve
the resolution of lens.

· YUV super resolution:

For full image composition, outputting images with better clearness and contrast, reduce noise and increase the equivalent number of pixels of the lens, for example, input 8MP image and output 32 MP image.

· ZOOM Super Resolution：

Optimized for continuous zoom factor, output images with better clearness and contrast, reduce noise

Improve the clearness of telephoto lens’ optical zoom-in effect, making the camera more capable of telephoto shooting.

Panoramic Imaging

The panorama imaging algorithm captures frames in real time during the rotation and movement of the mobile camera, calibrating and splicing each frame according to the order to achieve ultra-wide-angle shooting, this helps users to capture scenes that cannot be even met by wide-angle cameras.

· Real-time preview of composition, you get what you see

· Support up to 360 degree capturing from different directions.

· Support without gyroscope assist

Ultra Wide-angle Distortion Correction

Due to the inherent characteristics of the ultra-wide-angle lens, when the user uses the mobile phone to preview or take pictures, the edges may be deformed and distorted. The distortion correction algorithm obtains data from lens’ calibration data to correct the photo distortion to the acceptable range of the human eye.

· Support preview, shoot and video after distortion correction

· Support zooming of ultra-wide-angle

· Edge portrait protection

Wide Color Spectrum Purple Fringes Correction

When the camera is shooting, due to the large contrast between bright and dark sides of the subject, purple fringing appears at the junction of high light and low light, usually in buildings, leaves and tree trunks, and the edges of lights.

The algorithm removes the optical purple fringes while preserving the pixels of the captured subject, avoiding the removal of effective image pixels.

· In addition to remove purple fringing, the algorithm can also remove the resulting yellow, green, and red fringing

· Retain the origin edges of purple and blue objects, which avoids mistaken removal

Filters

By modifying certain parameters, linear processing is performed on each RGB channel of the image, so as to add a unique color scheme to the image. Currently, a total of 100+ types of colorful filters are provided, supporting preview, captures and video. Later, through color mapping, you can obtain a high-precision reference for the third party filter effect.

· Support filter effect customization

· Support manual adjustment of filter density

Intelligent Scene Recognition

Simultaneously analyze 1000+ different categories of texture, color, brightness, and structure information through a deep neural network, it accurately identifies scenes in images and video, and provides accurate scene information for optimizing ISP parameters and intelligently switch post-processing modes (portrait mode, smart filter).

· Support recognition of sunrise and sunset, night, green plants, landscape, flowers, sky, food, fireworks, text, cat, dog, portrait, sea, building, grass and moon

· Extremely fast, stable, efficient, support preview

AI DSLR Mode

By collecting a large amount of beautified photos in different scenes, training the model via deep learning. The AI model can adjust the brightness, color, saturation, contrast and sharpness of different areas in the image, so that each part of the photo reveals in suitable effects and the whole picture is more ornamental.

· Supported scenes: sunrise and sunset, green plants, landscapes, flowers, blue sky, food, text, cats, dogs, sea, buildings

· Support preview and photo taking

· Avoid ISP tuning for different scenario

· Automatic AI beautification for different content in the scene

Trapezoidal Document Correction

In scenes such as classes, conferences, exhibitions, etc., it is necessary to snapshot blackboards, documents, projections, display screens and so on. However due to the limitation of the angle, the image containing the content is tilted trapezoid. This algorithm corrects the trapezoidal content.

· Supports edge recognition and correction for documents, blackboards, monitors, projections, etc.

· Support manual adjustment of edges

Shadow Removal in Documents

There is often a block of light while shooting papers in the room and as consequence shadow appears in the photo. This algorithm removes shadow and enhances words to produce more clear photos via increasing the contrast.

· Remove shadow produced by blocking light

· Increase the clearness of words

· Consistent color of words

AI Single Take

Single take technique based on AI can take different photos and videos to record moments at the time, and automatically process photos and videos with different effects, choose the best moment and share it on social media.

· Based on the content and capturing conditions，

taking videos and processing different fragments in different effects.

· Based on the content and capturing conditions，snapping wonderful moment automatically

· Based on the content，editing smartly

· Based on the content，adaptively using filters or other algorithms like portrait bokeh

Video Stabilization

Based on the real-time motion data collected by the gyroscope, the algorithm analyzes each frame, establishes the feature points, and compensates the picture reversely to get a stable video.

· Support 6 degrees of freedom stabilization

· Shear rate can be adjusted based on stabilization, the higher shear rate, the better stabilization.

· Support rolling shutter distortion correction

Black Sesame Technologies Co., Ltd
Media information

pr@bst.ai
Business cooperation:

mkt@bst.ai
Send us a resume:

recruiting@bst.ai

COMPANY Basic Information Our Strengths Company Culture Cooperation
TECHNOLOGY IP Core SoC Design Shanhai Toolchain
PRODUCTS Huashan Series Chips Wudang Series Chips Hanhai Middleware
SOLUTIONS Driving Solutions V2X Solution Industry Applications Consumer Electronics
NEWS News Center
INVESTOR Listing Documents Financial Reports Announcements & Notices Shareholder Information Corporate Governance IR Contact
CONTACT US
Contact

Wuhan

+

-

Address: 32nd Floor, Shenzhen State Investment Center, 1278 Heping Avenue, Qingshan District, Wuhan

Telephone: 027-86434706

Shanghai

+

-

Address: NO.2, Lane 358 Jinhu Road, Office Park, Pudong New Area, Shanghai

Telephone: 021-58380278

Beijing

+

-

Address: 4th floor, Kejian Building, Building 6, Yard 1, Zhongguancun East Road, Haidian District, Beijing

Shenzhen

+

-

Address: 32nd Floor, Block B1, Building 12,Shenzhen Bay ECO-Technology Park, No.18, South Keji Road, Gaoxin Community, Yuehai Street,Nanshan

Telephone: 0755-86952165

Chengdu

+

-

Address: Floor 3, Unit 1, Building 4, No. 219 Tianhua Second Road, Chengdu High-tech Zone, Sichuan Province

Telephone: 028-81198576

Chongqing

+

-

Address: 21th Floor, Building B11,No.9 XianTao Data Valley Middle Road,YuBei District， ChongQing

Telephone: 023-67451527

America

+

-

Address: 2290 North First Street, STE100，San Jose, CA 95131

Telephone: +1 669-9008311

Singapore

+

-

Address: 01 Fusionopolis Walk #04-05/06 Solaris South Tower Singapore 138628

Telephone: +0065 69082986