Best algorithms of DVS and RGB lens

The AI Vision Technology.

Let us introduce you the most advanced visual technology.
You will see the AI processing effect of DVS lens & RGB lens.
Let your infinite imagination come into play!
Compare to RGB Lens, more detail features can be shown by DVS Lens and not affected by illumination.


The best choice In-vehicle autopilot L4, L5.

Compare to the RGB lens,
the DVS has the main features:
– Event-base vision feature
– High-speed
– Low-latency response
– Wide dynamic range operation –  Focal-plane data compression
– Low consumption

As shown in the figure, we can see with the naked eye that in the same scene, DVS can see much more details than RGB.
This is a revolutionary product in vision AI.

Privacy protection

DVS can do what RGB can!!!

In addition to the identity identification, the DVS can do everything that the RGB can do, and even more efficient, more accurate, because the lens is not affected by illumination, and can even protect your privacy.

The Advance Heath care

Any posture or gesture, we can always find out quickly!

In the health care algorithm introduced in the previous paragraph, it was implemented using RGB lenses,
but DVS lenses can also be the same as RGB lenses, or even more.,
We can even say that more efficient, more accurate,
because the lens is not affected by illumination, and can even protect your privacy.

The 3D mapping

We also use in sweep vacuum robot.

We can clearly see the edge and all the details of the scenarios,
and even the imperceptible liquid can be expressed! 
This is very suitable for 3D mapping or AR field etc.
It can even detect the distance accurately through dual DVS lens.


Complex Facial Expression Recognition

For RGB lens use, we have the “Exemplar-Embed Complex Matrix Factorization for Facial Expression Recognition”.
This algorithm has more efficiency with competitors and can even recognize up to 7 emotions, like surprise, sadness, anger, fear, disgust, happiness and neutral, and it can even recognize you AGE/Gender.

The Accuracy is as high as 94.3%

Driver Behavior Detection

Emotion, drowsiness, talking on the phone and smoking, etc.

Dangerous driving is terrible!
Talking on the phone, smoking and dozing off , etc. while driving are all very bad.
In addition to detecting whether the driver has these behaviors, our technology can further detect whether he is drowsy.
As shown in this demonstration, we can accurately identify the driver even if he is wearing glasses or sunglasses.
We can even combine emotion detection to automatically play/suggest music that you can listen to, etc.

The Objects.

Objects recognition & Feature extractive

No matter general or medical objects, our technology always accurately recognized,
Even you can freely recognize the specific targets.

The Accuracy is as high as 93%

Different feature extraction

Dual-RGB complexity feature identification.

Compare to general feature extraction, it’s vary suitable for defect detection with high yield and high complexity feature recognize requirement.
Also, defect types and classifications can be tracked in real time.
In addition to new defect type can automatically rebuild the model.
This algorithm can even accurately calculate the deviation value of the pin height.
The Accuracy is as high as 97%



The RGB Lens counting

We can do more in addition to counting!!!

In addition to the AI technology that you generally know about counting the number of people from the top to the bottom of the lens,
What we are demonstrating here is a lens from a general angle on the side of the road.
Through our AI technology, the number of people can also be counted,
and even the same thing can be done with the DVS lens with privacy protection.

The Accuracy is as high as 97%

The RGB Lens grouping

Grouping analysis to make prediction.

After finding the group,
we can analyze its behavior,
further predict the next action, and use it to prevent danger.

The Accuracy is as high as 92%

Cargo containers number identification

AI technology different from license plate recognition

In addition to identifying the license plate, we can also identify the number of the cargo container.
Due to the different positions of the container numbers,
it is more difficult to find the container numbers.
our technology can still find and identify the correct characteristic value of the container number from different locations and a variety of information.

The Accuracy is as high as 93%

Advance Gestures

Multi-mode gesture detection & identification

How to define the “Multi-mode”?
We can say that no matter you use your gestures through swipe or fixed-posture, even compound gestures.
Our algorithm can process in real time and is very lightweight,
it can recognize a variety of complex gesture recognition.
You can even apply this algorithm to the DVS lens, If you need to protect your privacy.

Heath care

Any posture or gesture, we can always find out quickly!

Health care is very important.
Our vision AI “Dynamic Tracking Attention Model ” is very comprehensive,
as if your child is about to cross the danger zone or make dangerous actions, we can all recognize it.
We can also accurately identify your actions. What’s more worth mentioning is that we are suitable for fall down detection in a variety of different postures.

In health care field, we can do such as “Gesture distress signal detection”, “Predict posture risk behaviors(domestic violence, cordon, etc.)”, “Visual intrusion detection”, “Sitter-in” and “Baby care”, etc.
In addition, we are suitable for fall down detection in a variety of different postures.

Lip language recognition.
Fall, Baby care.

Stroke disease

CT-base Cerebrovascular analysis

Our technology can automatically determine the images produced by computed tomography,
such as changes in cerebral vascular diseases and predict pathological trends,
it can assist doctors in rapid diagnosis and treatment.
thereby saving medical time and reducing misjudgment possibility.

The Fracture analysis

Fracture detection and type judgment.

Through the X-ray image of the bone scan, we can quickly identify and accurately list the suspicious abnormal features in advance,
so that the doctor can directly target these targets to determine whether the abnormality is true and further treatment,
thereby saving medical time and reducing misjudgment possibility.

The disease trend prediction

RNA-base Cancer classification and disease trend prediction & Establishment of vascular structure.

In addition to Establishment of vascular structure, it was including Breast, lung, brain, blood, prostate, kidney, bowel, stomach, head & neck, bladder, ovary, thyroid, uterus, liver, skin, pancreas, soft_tissue, cervix, testis.
From top 20 TCGA dataset in cBio (except PNS).

Vascular structure establishment