Amazon Go Introduction

Amazon Go Project Analysis

What's is Amazon Go?

Amazon Go is a new kind of grocery store with no checkout required. Amazon proposed the Checkout-free Shopping experience. With the Amazon Go app to enter the store, grab the products you want, and just go! No line, no checkout.

 December 5, 2016, Amazon publish a video about the Amazon Go.
Four years ago, we started to wonder, what would shopping look like if you could walk into a store, grab what you want, and just go? What if we could weave the most advanced machine learning, computer vision and AI into the very fabric of a store, so you never have to wait in line? No lines, no checkouts, no registers. … Use the Amazon Go app to enter, then put away your phone and start shopping.
Take whatever you like. Anything you pick up is automatically added to your virtual cart. If you change your mind about that cupcake, just put it back. Our technology will update your virtual cart automatically.
So how does it work? We used computer vision, deep learning algorithms and sensor fusion, much like you’d find in self-driving cars. We call it “Just Walk Out” technology. Once you’ve got everything you want, you can just go. When you leave, our “Just Walk Out” technology adds up your virtual cart and charges your Amazon account. Your receipt is sent straight to the app.

How does it work?

The checkout-free shopping experience is made by the same technology used in self-driving cars: computer vision, sensor fusion, and deep learning. The checkout-free technology can detect when the product is taken from or return to the shelf, and keep track of them in your virtual cart. When you are done the shopping and walk out the store, the technology will charge your amazon account and send you a receipt. 

 Some technique details:
1 Train good model with Big Data: Since the store can collect huge sets of data include videos, movement sensors, RFID trackers, and other data. Such big data can be used to train some high-performance deep learning models. 
2 Improve the products displays: The artificial intelligence can learn from customer flows, and figure out how to display the products at which place to increase the card size and maximize sales.
3 Personalization services and offerings: Having the customer data both online and in-store, the amazon can fine-tune at a granular level to serve a customer. 
4 Detect products with RFID: The RFID is used to detect when a product is taken from or return to the shelf and syncs the data with the online virtual cart.

What can I buy at Amazon Go?

The Amazon Go store provide the ready-to-eat breakfast, lunch, dinner, and snack options make fresh every day by the Amazon chefs and favorite local kitchens and bakeries. 

The Challenge Problem:

1 Twins problem: if there are twins walking into the store, how the walk out technology can differentiate them?
Walk out with scanning the ID
2 If there a big group of people entering the store, can the walk out technology distinguish them?
More cameras and higher computation capability
3 Comparable with an alternative design: Just use the RFID, let the custom to walk out some region one by one. The scanner can detect all the products in one second and charge the amazon account. Then we don't need the computer vision and deep learning technology. 
Still need checkout

 Future blog topic:
1 Identity verification (Face vs body)
2 How the amazon label the data?
3 CNN paper reviews
4 color image vs the depth data






Comments

  1. gallingabrielliMarch 4, 2022 at 6:37 PM

    Borgata Hotel Casino & Spa Launches New $1.2B Atlantic City
    The Borgata Hotel 제주도 출장마사지 Casino & Spa has announced the launch of 전주 출장마사지 its $1.2 billion expansion 의정부 출장마사지 project 구리 출장안마 on Atlantic City's Boardwalk, 상주 출장샵

    ReplyDelete

Post a Comment

Popular posts from this blog

The difference between autoencoders and a restricted Boltzmann machine?

Image
Autoencoder is a simple 3-layer neural network where output units are directly connected back to input units. E.g. in a network like this: output[i] has edge back to input[j] for every hidden node. Usually, the number of hidden units is much less than the number of visible ones. As a result, when you pass data through such a network, it first encodes input vector to fit a smaller representation and then decodes it back. The task of the training is to minimize an error of reconstruction or find the most efficient compact representation for input data. The AEs (Autoencoders) are similar to PCA and can be used to dimension reduction. The hidden layer decoded features can be used as input features for downstream classification or another AE. RBM  is a  generative   artificial neural network  that can learn a  probability distribution  over a set of inputs.  RBMs are a variant of  Boltzmann machines , with the restriction that...
Read more

SSD: Single Shot MultiBox Detector

Image
SSD is the extension or updating version of YOLO. Compared with YOLO: 1 SSD use the multi-scale feature maps for detection, but the YOLO only use one scale for detection. 2 Before the decision layer, the YOLO use the fully connected layer for detection, but the SSD use the convolution filter for detection, which can improve the final detection performance. 3 Default boxes and aspect ratios. there are a set of default bounding boxes with each feature map cell for multiple feature maps at the top of the network. R-CNN: Selective Search Region Proposals; SVM classification Fast R-CNN: Selective Search Region Proposals; Shared computation by using SSP. Faster R-CNN: Region Proposal Network OverFeat VS SSD: If the SSD only use one default box per location from the topmost feature map, the SSD would have similar architecture to OverFeat. YOLO VS SSD: If use the whole topmost feature map and add a fully-connected layer for predictions instead of convolutional predictors, it...
Read more

You Only Look Once

Image
Title: You Only Look Once: Unified, Real-Time Object Detection YOLO is a new approach for object detection. You only need run a single CNN on the image once and make the decision for each grid. Instead of using the sliding windows or region proposes. the YOLO proposed to divide the image to S*S grid and predict each grid with regression. Contribution: The traditional model will do a classification for each proposal or sliding window, but the proposed YOLO divide the image to S*S grid and train a regression model. The ground truth target vector has the dimension of S*S*(B*5+C). Each grid cell predicts B bounding boxes and confidence scores for those boxes. The confidence score reflects how confidence the current grid contain the object and also how accurate it thinks the box is that it predicts. Network: The network has 24 constitutional layers followed by 2 fully connected layers. There are four 2*2 max-pooling layers with stride 2. Instead of inception modules used by ...
Read more