Complete How To Guide
ATLAS VISION turns any camera -- your phone, laptop, or webcam -- into a superhuman eye with 14 different vision powers.
It can detect objects, recognize faces, read sign language, measure depth, track human bodies in 3D, identify products, read emotions, and much more -- all in real time.
Think of it like giving your camera a brain that understands everything it sees.
Getting started takes about 10 seconds.
Your camera is how ATLAS VISION sees the world. Here's how to turn it on.
Tip: If camera access was blocked by mistake, go to your device's Settings, find your browser app, and turn Camera back on.
Object Detection is the most basic superpower. When you point your camera at something, the AI draws a colored box around every object it recognizes.
Above each box, you'll see a label -- like "person," "car," "bottle," or "dog." Next to the label is a percentage that shows how confident the AI is. For example, "car 97%" means the AI is 97% sure that's a car.
It can see hundreds of objects at once. Point at a busy street and you'll see boxes around every person, car, traffic light, and backpack -- all labeled in real time.
This uses a model called YOLO (You Only Look Once), which is one of the fastest object detectors in the world. It processes video frames so quickly it feels instant.
Segmentation goes beyond just drawing a box around something. Instead of a simple rectangle, it traces the exact outline of every object, pixel by pixel.
Imagine coloring in an object with a highlighter pen, perfectly following its edges. That's segmentation. It colors in every single pixel that belongs to each object.
ATLAS VISION uses SAM4D (Segment Anything Model in 4D), which tracks objects across video frames over time -- not just in a single photo. It can even fuse camera data with LiDAR depth sensors on newer iPhones and iPads for military-grade tracking accuracy.
Each tracked object gets its own ID, so the AI remembers "that's the same dog" even if it walks behind a tree and comes back into view.
Exploded View is like taking something apart to see what's inside -- without actually touching it.
Point your camera at any object, and the AI will figure out all the individual parts and components that make it up. It's like an X-ray that breaks things down into pieces.
For example, point it at a bicycle and it might show you: frame, wheels, chain, pedals, handlebars, seat, brakes, and gears -- each one labeled and highlighted separately.
Engineers, mechanics, and curious people love this mode because it helps you understand how complex objects are built.
Text-to-CAD lets you describe a 3D object in plain English and the AI will generate a 3D engineering model from your words.
Just type something like "a gear with 24 teeth and a 10mm bore" or "a phone stand with an adjustable angle," and the AI creates a 3D blueprint you can view, rotate, and export.
Combined with Exploded View, you can point your camera at something, see all its parts, and then ask the AI to recreate any part as a 3D model. This is powerful for prototyping, repair, and manufacturing.
Body4D gives the AI a complete understanding of the human body in 3D space. It doesn't just see "a person" -- it maps out the entire skeleton with 70 joints, including every finger, elbow, knee, and head position.
It creates a full 3D mesh of the human body -- like a digital mannequin that matches your exact pose and body shape in real time.
Even if part of your body is hidden behind a chair or table, the AI uses occlusion recovery to figure out where your hidden arm or leg probably is.
This works across video, so it tracks your movement over time (the "4D" part -- 3D space plus time). Great for fitness tracking, sports analysis, physical therapy, and dance.
Depth Perception lets the AI understand how far away things are -- just from a regular camera image.
It creates a "depth map" where close objects appear bright and far-away objects appear dark. This tells the AI the 3D layout of the entire scene.
On newer iPhones and iPads with LiDAR sensors, it can give you exact distances in meters by fusing AI estimation with real laser measurements.
Use cases: measuring room dimensions, understanding spatial relationships between objects, parking assistance, and making the other vision features (like body tracking) more accurate.
Face Recognition identifies WHO a person is, not just that a person is there. It uses a technology called ArcFace which creates a unique 512-number fingerprint for each face -- far more accurate than older systems.
Once a face is registered, the AI will recognize that person again in the future, even from different angles or in different lighting.
Emotion Detection reads facial expressions and classifies them into 7 emotions: happy, sad, angry, surprised, fearful, disgusted, or neutral. It runs on every frame that contains a face, so it updates in real time as expressions change.
This is useful for security, smart home greetings, accessibility tools, and understanding audience reactions.
Sign Language Recognition lets ATLAS VISION understand American Sign Language (ASL) in real time.
Point your camera at someone signing, and the AI will detect their hands, recognize each letter or sign, and convert it to text on screen. It supports all 26 letters plus common word-level signs.
It works in two modes: single-frame (recognizing one letter at a time) and continuous (reading a sequence of signs to form words and sentences).
This makes communication accessible for deaf and hard-of-hearing people -- anyone can understand sign language through the camera.
Item Recognition goes beyond basic detection. When the AI sees an object, it enriches it with detailed product information.
It uses the UNSPSC industrial classification system -- the same system used by governments and big companies to categorize every product in the world with an 8-digit code.
Point your camera at a laptop, and instead of just "laptop," you might see: "Notebook computer (43211503) -- Information Technology > Computers > Notebooks."
It can also search the web to find the exact brand, model, and price of products. Great for inventory, shopping comparison, and warehouse management.
Scene Classification looks at the big picture instead of individual objects. It tells you what kind of place you're looking at.
Point your camera around a room and the AI will say: "kitchen," "office," "gym," or "parking lot." It understands hundreds of scene types.
This helps the other AI features work better. For example, if the scene is a "kitchen," the AI knows that rectangular objects are more likely cutting boards than laptops.
Food Detection recognizes what food is on your plate and can estimate nutritional information like calories, protein, carbs, and fat.
Point your camera at a meal and the AI will identify each item -- "grilled chicken breast," "brown rice," "steamed broccoli" -- and provide an estimated nutrition breakdown.
Perfect for diet tracking, meal planning, and health-conscious eating without having to search for every food item manually.
Visual Translation reads text in the camera view and translates it into your language in real time.
Point your camera at a sign, menu, document, or label in another language, and the AI will detect the text, figure out what language it's in, and show you the translation overlaid right on the screen.
This is incredibly useful when traveling, reading foreign product labels, or working with documents in unfamiliar languages.
Medical Visualization applies the AI's vision capabilities to health and medical contexts.
It can analyze skin conditions, highlight anatomical landmarks on the body using Body4D pose data, and assist with understanding medical images.
Combined with depth perception and segmentation, it can create detailed 3D views of body regions that help healthcare professionals and patients alike.
Note: This is an assistive tool only. Always consult a real doctor for medical decisions.
SHANGO Night Vision is the advanced security module. It enhances low-light and near-darkness footage so the AI can still detect objects, faces, and movement when regular cameras see almost nothing.
It uses AI-powered image enhancement to brighten, sharpen, and de-noise dark frames, then runs the full detection pipeline on the enhanced image.
This is designed for security monitoring, outdoor surveillance, and any situation where you need vision after dark. Combined with depth perception and face recognition, SHANGO provides 24/7 intelligent awareness.
SHANGO capabilities are part of the advanced security tier.
ATLAS VISION organizes its powers into modes you can switch between instantly.
How to switch modes: Tap the mode buttons at the bottom of the screen. The active mode will be highlighted in gold. You can switch at any time -- the view updates instantly. Each mode activates different combinations of the 14 capabilities listed above.
Follow these simple tips and ATLAS VISION will work at its best.