Deliver Superior AI-Powered Smart Retail Solutions Faster

Develop Superior Smart Retail Applications with Accelerated Inference Performance

The Challenge

Navigating the delicate balance between accuracy and latency is crucial for your smart retail solutions. High accuracy is paramount, but must be paired with low latency for real-time responses. Furthermore, cost-efficient scaling of inference is essential to meet varying demand without sacrificing performance. Any imbalance can disrupt the user experience, causing delays or inaccuracies. Your challenge lies in harmonizing these elements for smooth shopper interactions.

Cut Your Inference Cloud Cost

Eliminate up to 80% of your compute processing cost by improving your models' inference efficiency.

Maximize Hardware Utilization

Make the most of your devices and scale up inference more cost efficiently with better hardware utilization.

Improve Product Quality with Better Accuracy

Improve model accuracy to prevent false alerts and deliver superior solutions

Improve Shopper Experience with Real Time Insights

Boost inference speed and achieve real-time performance without compromising on accuracy.

See It in Action

Watch a compilation of clips demonstrating how you can use Deci in a variety of retail use cases

Live Coding: How ResNet Changed Computer Vision

ResNet is an artificial neural network. It is a gateless or open-gated variant of the HighwayNet, the first working very deep feedforward neural network with hundreds of layers. In this walkthrough, Harpreet Sahota, DevRel Manager at Deci, demonstrates ResNet in action. You can check out the notebook here and follow along!

Watch Now

Live Coding: The EfficientNet family: Power, Speed, Accuracy and Efficiency

EfficientNet is a convolutional neural network architecture and scaling method that uniformly scales all dimensions of depth/width/resolution using a compound coefficient. Join Harpreet Sahota, DevRel Manager at Deci, for a walkthrough of ResNet in action. You can check out the notebook and follow along!

Watch Now

Live Coding: RegNet- The Most Flexible Network Architecture For Computer Vision

RegNet is a highly flexible network design space defined by a quantized linear function. It can be specified and scaled for high efficiency or high accuracy. Join Harpreet Sahota, DevRel Manager at Deci, for a walkthrough of RegNet in action, using it to determine whether an image is Santa or not. You can check out the notebook and follow along!

Watch Tutorial

Get Similar Results for Your Specific Use Case

Enabling Real-Time Semantic Segmentation for an Automotive Application

An automotive company running a U-Net based segmentation model on a NVIDIA Jetson Xavier NX struggled to achieve the target latency in production.

Using Deci’s AutoNAC engine, a faster and smaller model was generated. Latency was reduced by 2.1x, model size was reduced by 3x, and memory footprint was reduced by 3x – all while maintaining the original accuracy.

Real-time Depth Estimation
for an ADAS Solution

A large company delivering solutions for ADAS/AD needed to improve latency performance of a segmentation model used for real-time depth estimation for an automotive use case. The original model was based on MobileNetV2.

The company’s team used the Deci AutoNAC engine to build a custom model architecture to meet their inference performance targets while maintaining their original accuracy on the NVIDIA RTX 2080Ti hardware. As a result of this, the company is now able to deliver real-time inference performance for its automotive application.

Enabling Real-Time Classification
at the Edge

An automotive company wanted to improve the latency of a model powering their road condition estimation system. The poor latency of their baseline image classification model was impacting the real-time performance on their target hardware, a NVIDIA Xavier AGX.

The team used Deci’s AutoNAC engine to build a customized model, resulting in a 2.1x acceleration in latency. Accuracy was improved by +4%, throughput was improved by 2x, and the model size was reduced by 3.2x.

Application Showcase

Deci is used by retailers and smart retail solution providers to develop and optimize a wide range of applications. Here are some examples.

Self Checkout

Shelf Replenishment

Theft Prevention

Product Discovery

Supply Chain Management

Waste Management

Build Better Models Faster with Deci’s Deep Learning Development Platform

Main Capabilities Overview

The Deci platform is used by data scientists and machine learning engineers to build, optimize, and deploy highly accurate and efficient models to production. Teams can easily develop production grade models and gain unparalleled accuracy and speed tailored for any performance targets and hardware environment. Deci is powered by AutoNAC (Automated Neural Architecture Construction), the most advanced and commercially scalable Neural Architecture Search engine in the market. AutoNAC performs a multi-constraints search to find the architecture that delivers the highest accuracy for any performance targets and hardware environment.

Gain Superior Performance with Custom Architectures

Build accurate & efficient architectures tailored to your hardware and application’s performance targets with Deci’s Neural Architecture Search engine.

Simplify Runtime Optimization

Easily compile and quantize your models (FP16/INT8) and evaluate different production settings with a click of a button.

Maximize Accuracy with Advanced Training Techniques

Train models with SuperGradients. Leverage custom recipes and advanced training techniques (e.g. knowledge distillation, quantization-aware training) with one line of code.

Streamline Deployment with 3 Lines of Code

Deploy your models with Infery, Deci’s simple-to-use, unified, model inference API. Streamline deployment and boost serving performance with parallelism and concurrent execution. Compatible with multiple frameworks and hardware.