Home Bots & Business Blog: How to capture product data from price tags with OCR

Blog: How to capture product data from price tags with OCR

by Guest

For many retailers, competitive pricing analysis is the key to increase their revenue and stay ahead of competition. Global shifts are forcing these companies to become more data-centric, collect and analyze pricing data, map their position against competitors, and offer optimal prices.

Collecting the right data, however, can be difficult, especially in-stores. Not all stores have online shops to crawl and even when they do, prices can vary significantly in physical stores, especially when they are geographically far apart.

That’s why retail companies pay a lot for sales representatives visiting stores and manually writing down the prices. It’s a time-consuming and error-prone job which has to be carried out at least once a month.

Wouldn’t it be easier to simply scan the shelfs with your smartphone and automatically capture data from price tags within seconds? In this blog, we will explain how you can do this with the help of OCR, and what you need to get started.

What is Optical Character Recognition (OCR)?

Let’s start with the basics. What is OCR? If you already know, you can skip this part, but we would like to make sure that we are on the same page.

OCR stands for Optical Character Recognition and is a widely used technology to recognize text on printed or handwritten documents and images. It is used to convert virtually any document containing text (physical or digital, images, PDF files, emails, and so on) into machine-readable text data (e.g. XML, CSV, JSON).

OCR technology has proven immensely useful in automating workflows. It helps people work faster with minimal errors in tasks like data extraction, data entry and document verification.

Now that we have covered the basics, let’s have a look at how OCR can be applied to price tags.

How does OCR work on price tags?

In order to extract data from price tags, the first thing you need is a scanner, smartphone or other hardware device that can capture an image and allows the identification of characters. Usually a smartphone is used in combination with an OCR application.

If you have an application yourself or want to build one, the easiest way to enrich it with OCR capabilities is to integrate an OCR SDK into the application.

SDK stands for software development kit. Also known as a devkit, an SDK is a set of software-building tools for a specific platform to create applications. It includes a range of resources in its set of tools, including code libraries, documentation, code samples, debuggers, and testing and analytics tools.

The biggest advantage of SDKs is that they make developers’ jobs easier. Developers don’t have to build all the components of an app themselves, but they can “use” the work of others, saving them loads of developing time and money.

We can imagine this sounds a bit technical, so let’s describe the steps in the OCR process one by one.

The steps of the OCR process

The OCR process consists of several steps, starting with uploading images or PDF files and ending with exporting the recognition results.

  • Uploading image or PDF file
  • Pre-processing the image
  • Image to text conversion
  • Conversion to structured output
  1. Uploading image or PDF file

The first step is to provide a picture or PDF file of a price tag to the SDK. This is very easy, because the SDK allows you to detect custom objects.

You can simply scan a shelf with your smartphone and when the SDK detects a price tag, it draws a bounding box around it confirming that it detected a price tag. The SDK also features real-time feedback. It indicates, for example, if the user is too far away from the shelf, or if the environment is too dark to take a good picture.

The camera can automatically take a picture if it determines that the image quality is good enough.

  1. Pre-processing the image

After taking the picture, the image quality is enhanced during the pre-processing step to increase the recognition accuracy. The SDK optimizes brightness, enhances grayscale, detects the edges of the price tag and crops it from the background.

  1. Image to text conversion

Once the cropped image is received by the OCR SDK, it extracts the data from the price tag. At this stage, the price tag is analyzed for light and dark patterns, where the dark areas are identified as characters that need to be recognized and the light areas are identified as background. The dark patterns are then further analyzed to find alphabetic letters or numeric digits.

When all text is identified, natural language processing (NLP) helps to determine which text has what meaning. In this way, the SDK is able to determine what text belongs to the description, what numbers belong to the prices, and so on.

  1. Structured output

The SDK offers many options for exporting recognition results. Just having a plain text file is not as useful for pricing analyses. That’s why the OCR takes the text and converts it into structured output, for example in a JSON format. This is a format commonly used for transmitting data in applications.

From there, processing the data into your database or other software is super easy.

What data can you extract from price tags?

With the OCR process described above, you can capture an image of a shelf, crop out the price tags, and extract the prices. But price tags can contain a lot more information. Fortunately, OCR software is not limited to prices, it can extract other data points as well:

  • Merchant name
  • Product description
  • Product category
  • Volume price (e.g. price per kg / liter)
  • Product price
  • Serial number or barcode
  • Discount percentage and price
  • Date

What are the advantages of OCR?

Although there are numerous benefits of OCR, it mainly helps businesses in increasing the effectiveness and efficiency of the work. Its ability to quickly extract data from documents and objects is extremely useful, particularly in retail settings, which deal with high-volume scanning.

Following are some of the major advantages of OCR:

  • Saves up to 70% in processing time
  • Reduces costs by 40% to 75%
  • Improves productivity
  • Minimizes errors
  • Besides that, collecting data from price tags has two clear advantages for retail companies:
  • Enables a deeper understanding of the market
  • Improves marketing and sales strategies
  • Saves up to 70% in processing time

As opposed to traditional data collection, the use of OCR allows you to speed up the process dramatically. When done manually, sales representatives need to check price tags, note down prices and enter them into a system. On average, automating this process can save up to 70% in processing time.

The logic behind it is quite simple: technologies like OCR extract data from documents almost instantly. What was previously a slow and repetitive data collection process, is now done within seconds.

Reduces costs by 40% to 75%

Time is money, so saving time saves you a lot of money as well. There’s no magic formula as every company has different needs, but intelligent automation typically results in cost savings of 40% to 75%, with payback periods ranging from several months to years.

Improves productivity

OCR software helps you to improve your business productivity by facilitating quicker data collection. The time and effort employees were required to put in for extracting relevant data can now be used to work on core activities. As a nice little bonus, your employees will be happier as they can spend their time on more value-adding tasks.

Minimizes errors

Everyone makes mistakes at work. Whether it’s forwarding an email to the wrong person, entering the wrong order number, or shipping the wrong product, human errors happen. This is especially true for slow and labor-intensive processes like manual data entry. Error rates typically range from 0.55% to 3.6%, with outliers to as high as 26.9%.

Fortunately, automated data entry using OCR significantly lowers these rates by eliminating the risk of distractions, typographical errors and other mistakes commonly found in manual data entry. This results in better, more accurate pricing data that can be used to make well-informed business decisions.

Enables a deeper understanding of the market

Collecting accurate pricing information from price tags helps you to better understand consumer behavior. It also helps to identify how you position yourself in the market.

The more information you get from customers and products, the more you can adjust your business to their needs. It will become much easier to understand what customers are looking for and what they are willing to pay for your products.

Improves marketing and sales strategies

Having a better understanding of the market helps you to develop better marketing and sales strategies. This is particularly important when your products and your competitors’ products are sold in the same location.

If you notice, for example, that a competitor’s product is priced lower than yours and sales suffer from it, you can choose to run a promotional campaign for that product or give a small discount to recover the loss.

With current technological developments, there are no more excuses. It’s time to start automating and digitizing processes and gain that competitive advantage!

Yeelen Knegtering is CEO & Co-founder at Klippa

Misschien vind je deze berichten ook interessant