Cas d'utilisation

Solutions d'intelligence des données pour de meilleures décisions commerciales.

Des solutions puissantes pour moderniser la façon dont votre équipe tire des enseignements des données.


Relations publiques et communication


Les médias


Technologie juridique


Vision par ordinateur

La révolution de Sebring

La révolution de Sebring

L'industrie : Vision par ordinateur

Temps de déploiement : Lorem Ipsum


Le défi :

Sebring Revolution is transforming how we experience photography and cinematography with 3-D photography and virtual and augmented reality. Their studio captures subjects through the lenses of XXX cameras spanning angles from all 360 degrees. However, post production editing to remove sights of other cameras or doors in the studio from each frame is tedious, timely and expensive. Additionally the process of segmenting people from the background for use in Augmented Reality is also time consuming and expensive The team at Sebring Revolution outsources this work, which takes between five and seven days to complete, delaying their delivery considerably as well as costing them $0.20 – $3.50 per frame.

Objectif principal :

To improve the post-production process by decreasing the time to delivery and decreasing the cost to do so.

Solution sur mesure :

For this client we built automated processes to identify parts of images, cameras and doors, which need to be processed (object detection). This model automatically processes each image so that the small camera slots and doors are made invisible. The solution allows Sebring Revolution to own the entire editing process, speeding up their delivery by more than a week and cutting costs on outsourcing.

Modèles utilisés :

• Object Detection
• In-painting background
• Image Segmentation (subject from background)
• White Balancing
• Image Stabilization

Le processus :

Jour 1
Contrat signé
Jour 1 - 30
Onboarding (Methodology Review, Goal setting, Optimizing off the Shelf Models to solve needs)
Day 30 – 60
Data Gathering (Skim AI manages data annotator to generate labelled data to improve results)
Day 60 – 150
Training Custom Models(Segmentation, Object Detection, Stabilization and White Balancing)
La révolution de Sebring