OCR and customs operations: A basic tool in a complex world

In today's digital world, document management and processing have become more complex and important than ever. Given this complexity, how can we ensure that customs documents are processed efficiently and accurately? Is there a tool that can facilitate this process? And if so, is it really as powerful and sophisticated as is claimed?

One technology, Optical Character Recognition (OCR), is often cited as the solution. But OCR has its limits. Let's see where it can help, and where it may not.

‍

‍

What is OCR?

Before going any further, let's demystify what OCR is. OCR stands for Optical Character Recognition. It's a technology that transforms text from images - whether scanned documents, photos or PDF files - into editable, searchable data. Think of OCR as a very dedicated worker, who will read characters and transcribe them, without trying to understand their meaning.

The history of OCR

The concept of OCR, or Optical Character Recognition, is not new. Its origins lie in the 1910s, when Emanuel Goldberg, an inventor and scientist, developed a machine that read characters and converted them into telegraphs. However, it wasn't until the 1960s that OCR really became popular.

At that time, the American company Intelligent Machines Research Corporation (IMRC ) developed the first commercial optical character recognition system. The system, named "OCR-A", was able to read printed text and convert it into electronic data. The idea was to facilitate the digitization of newspapers and printed documents, making them accessible to a wider audience. This revolutionized the way information was stored and shared.

In the 1970s, OCR took a giant step forward with the introduction of font recognition technology. This made it possible to extend the use of OCR to a wider range of printed documents, including books and magazines.

Over the years, OCR has continued to develop and improve. It has evolved from simple text recognition to the ability to recognize and convert images, mathematical formulae, and even musical scores into digital data. What's more, OCR has been enhanced to recognize multiple languages and scripts, making the technology truly global.

Fun fact about OCR

Did you know that in 1980, The New York Times became one of the first newspapers to adopt OCR on a large scale? Kurzweil Computer Products, Inc. developed a specific machine to scan the newspaper's pages and convert them into editable text. This historic step marked the beginning of the digital era for print media, underlining the revolutionary scope of OCR.

Document processing with OCR

When it comes to customs operations, document processing can quickly become a tedious and time-consuming task. Shipping documents, invoices, customs declarations and much more have to be processed on a daily basis. When faced with such a large volume of data, OCR can be an invaluable tool.

But what does it really do?

OCR simply converts image text into editable text. It is not programmed to understand the data it processes, but simply to make it usable. Let's imagine a two-step process: firstly, a commercial invoice received by email or scanned, producing an image with text; secondly, OCR steps in to convert this image into editable text.

OCR therefore has no ability to understand the meaning of the characters it extracts, nor to predict how the converted data will be used in the future. For example, OCR cannot determine whether "1234" corresponds to an invoice number, a weight or a quantity, or whether "ABC Logistics" refers to a supplier, a product description or an Incoterm. Its main function is to transform these characters into a machine-readable format, without manual input.

For this reason, human intervention or other more advanced technologies are often required to process this newly digitized data, depending on the type of operation you wish to perform. This is where the combination of OCR and AI can truly revolutionize document processing, a subject we'll return to in more detail in a future article.

{{pop-up-component}}

‍

Data accuracy and OCR

One of the great strengths of OCR is its ability to increase data reliability. When data is entered manually, the risk of error is very high. These errors can lead to serious problems, such as delays in operations, penalties for non-compliance, and even disputes with suppliers or customers.

Thanks to OCR, this risk can be reduced. OCR technology is able to read and transcribe characters with great accuracy, ensuring that crucial information such as reference numbers, commodity codes and other important details are correctly captured.

‍

‍

The economic impact of manual errors and OCR intervention

Manual data entry is an error-prone process that can cost companies dearly. According to a Harvard Business School report, data entry errors cost over $3 trillion every year in the USA alone. And it's not just a matter of time: these errors can lead to a multitude of complications, such as incorrectly labeled goods, errors in order quantities, and even budgetary problems.

Eye strain, loss of attention and hard-to-read content can create a volatile cocktail of errors. Furthermore, training employees in manual data entry also represents a cost, both in terms of time and resources.

This is where OCR comes in, as an effective cost-saving solution. By automating the data entry process, OCR avoids many of these problems. Key information is captured automatically, limiting costly errors.

‍

‍

OCR limits

• Variable accuracy: OCR accuracy can be affected by a number of factors. For example, poor-quality images, complicated fonts or poorly structured documents can reduce the accuracy of OCR conversion.
• Character confusion: OCR can be confused by the nuances of certain characters. For example, a misplaced "1" may be confused with an "i" or an "l".
• Table processing: Import/export documents, for example, often contain complex tables with several values per cell, or information straddling several columns. OCRs can have great difficulty in discerning the different pieces of information and correctly aligning the data extracted from the tables.
• No understanding of context: OCR does not understand the context of the data it processes. It can't distinguish a supplier name from an invoice number, requiring additional technology or human intervention to correctly analyze and use the extracted data.
• Dependence on original document quality: As mentioned above, the quality of the original document has a direct impact on OCR efficiency. Poorly scanned documents, discolored documents or documents with superimposed characters can result in incorrect character detection.
• Difficulty with certain languages and scripts: OCR may have more difficulty with certain languages and scripts, especially those using certain alphabets. For example, Arabic or Chinese alphabets can pose significant challenges for OCR.

‍

‍

OCR: a tool, not a universal solution

Having examined the capabilities and limitations of OCR, one thing is clear: OCR is a valuable tool, but it is not a one-size-fits-all solution. It is an indispensable building block for automating the processing of documents in international trade, which are rarely digitized. It is the first building block of digital processing workflows which, combined with different technologies, have the capacity to transform the administrative tasks of this industry. OCR can help automate document processing, reduce input errors and speed up operations, but it is no substitute for human understanding or in-depth analysis provided by more sophisticated technologies such as AI.

Customs operations need more than just the content of documents. The declarant's expertise enables him to use the information contained in the documents to deduce the rest of the data required to complete his declaration. Extracting the content of documents alone is not enough to fully complete a customs declaration.

‍It's therefore essential to understand OCR for what it is: a tool that, used on its own, won't enable a high enough degree of automation for complex tasks, but when used correctly and combined with other technologies, it can bring considerable value to your customs operations.

‍