MUSEAI Smart Mannequin
A mannequin that restyles itself for every passer-by.
MUSE is an on-device computer-vision mannequin built on the NVIDIA Jetson Nano Developer Kit. It detects shoppers as they approach, analyses attributes and demographics at the edge, and changes the outfits and cross-sell suggestions it shows to suit each one. All inference happens on the device.
- Runs on
- NVIDIA Jetson Nano in the display
- Input
- Camera at the mannequin
- Output
- Dynamic outfits and cross-sell
- Privacy
- All inference on-device
Window displays that work for everyone who walks past.
A traditional mannequin shows one look to everyone, all day, until someone changes it by hand. It can't tell who is looking, and nobody knows whether it brought a single person into the store.
MUSE gives the display eyes and timing. A discreet camera and a Jetson Nano detect each passer-by, estimate attributes such as age band and gender, and pick the most relevant look and add-ons from your range to show on the screen beside the mannequin. It also counts how many people stopped and which looks held their attention.
What changes with MUSE
- One look for allThe most relevant look for each passer-by
- Manual changesLooks rotate automatically through the day
- No measurementPassers-by, stops and attention counted anonymously
- Missed add-onsCross-sell suggestions alongside every look
How MUSE works
From the moment someone walks into view to the moment the display changes, every step runs on the device.
Spot
A camera in the mannequin or display detects shoppers approaching and tracks them while they are in view.
Read the moment
On-device models estimate attributes and demographics, such as age band, gender and group size, on the Jetson Nano.
Pick the look
Your merchandising rules and a relevance model choose the outfit and cross-sell items most likely to interest that shopper.
Show and measure
The display updates to the chosen look and records anonymous counts: passers-by, stops, attention time and which looks performed.
Features
Everything a visual merchandiser would do for each shopper, if they could stand in the window all day.
Shopper detection
Detects and tracks people approaching the display in real time, on the device.
Attribute and demographic analysis
Estimates age band, gender and group size to choose relevant content.
Dynamic outfits
Shows the look most relevant to each passer-by, drawn from your current range.
Cross-sell suggestions
Pairs every look with add-ons such as footwear and accessories, and shows where to find them in store.
Attention measurement
Counts passers-by, how many stopped and how long they looked.
Display performance
See which looks drew attention by hour and day, and refine your visual merchandising.
On-device privacy
All inference runs on the Jetson Nano. No video or images leave the mannequin.
Merchandising control
Your team sets the looks, rules and schedules. MUSE chooses within the limits you define.
Works offline
Runs without internet, and connects only to sync content and anonymous counts.
Technical specifications
MUSE combines a camera, a Jetson Nano and a display with your mannequin. Everything it needs to decide runs locally.
NVIDIA Jetson Nano Developer Kit
- GPU
- 128-core NVIDIA Maxwell
- CPU
- Quad-core ARM Cortex-A57, 1.43 GHz
- Memory
- 4 GB 64-bit LPDDR4
- AI compute
- 472 GFLOPS (FP16)
- Power
- 5 W and 10 W modes
- Compute
- NVIDIA Jetson Nano Developer Kit, housed in the display
- Inference
- Fully on-device, with models optimised for the Jetson GPU using NVIDIA TensorRT
- Vision models
- Person detection and tracking, attribute and demographic estimation
- Selection logic
- Merchandising rules combined with a relevance model, per passer-by
- Camera
- Discreet camera integrated in the mannequin or display
- Display
- Screen beside or behind the mannequin, sized to the window or floor space
- Content
- Looks, product images, add-ons and in-store locations, synced from your range
- Measurement
- Anonymous passers-by, stops, attention time and look performance
- Connectivity
- Wi-Fi or Ethernet for content and count sync; detection works offline
- Data handling
- No video or images leave the device
A display that pays its way
MUSE gives merchandisers flexibility, stores more walk-ins, and leadership a way to measure the window.
Visual merchandising
- Run several looks from one mannequin
- See which looks stop people
- Change the display without re-dressing
Store teams
- More walk-ins from the window
- Add-on ideas that send shoppers to the right rack
- No content to update by hand each day
Brand and leadership
- Measure the window like a digital campaign
- Personalisation with no personal data collected
- A display people remember
Use cases
Anywhere a display is meant to stop people and bring them closer.
- Store windowsPull people in from the street or mall corridor with looks that suit them.
- Store entrancesGreet shoppers with the edit most relevant to them.
- Department-store floorsPromote collections across categories from a single display.
- Mall atriums and pop-upsBrand moments that measure their own reach.
- Seasonal launchesRotate new-season looks by time of day and audience.
- Sports and footwear storesPair apparel with the right shoes and accessories.
Questions about MUSE
Can't find your answer here? Ask us directly.
Does MUSE record shoppers?
No. Video is analysed in memory on the Jetson Nano and discarded. Only anonymous counts, such as passers-by and stops, are kept.
Does the mannequin change clothes by itself?
The mannequin wears a physical outfit. MUSE changes what the display beside it shows: alternative looks, colours and add-ons chosen for each passer-by.
Can we control which looks are shown?
Yes. Your team sets the looks, product rules and schedules. MUSE chooses within those limits for each shopper.
What does it need on site?
Power, a mounting point for the camera and display, and Wi-Fi or Ethernet for content sync.
Do MUSE and IRIS work together?
They complement each other. MUSE draws shoppers to a look in the window, and IRIS helps them try it and find what pairs with it at the mirror.
Make your window work harder.
Tell us about your store format, and we'll plan a MUSE display for it.