Building Production-Grade spaCy Text Classification Pipelines for Business Data
Unlock the full potential of spaCy with this guide to building production-grade text classification pipelines for business data.
Despite a challenging year for global logistics in 2020, investments in warehouse automation technologies witnessed a whopping 57% increase, hitting $381 million in the first quarter. Warehousing automation solutions drove deal activity in logistics as investors made huge bets on emerging robotics technologies for warehousing and supply chain.
Investors’ growing interest in startups dealing with warehouse automation coincides with the surge in ecommerce orders and changing consumer buying behavior. Smaller batch (single item) orders have become more popular post-pandemic and consumers now expect better visibility into the order fulfillment process.
It is no surprise then that warehouse automation is fast becoming a competitive advantage not only for ecommerce companies but all businesses reliant on efficient logistics.
This article will explore what warehouse automation is, what it entails, and what machine learning based software tools can push you forward in the world of warehouse automation.
Warehouse automation is a collection of technologies that enable organization logistics to orchestrate the movement of goods into, within, and out of warehouses to customers with minimal human intervention. Warehouses are ideal candidates for automation because they have multiple repeatable, process-oriented, and human error-prone operations. Think pallet stacking, order picking, receiving, doing inventory, etc. Automation helps warehouse processes become more consistent, accurate, and overall a profitable business unit.
The COVID-19 pandemic and its resulting impact on supply chain and logistics have served as a catalyst to warehouse transformation. With companies having to operate with decreased labor force, automation served a critical purpose. Conveyor systems and other robotic process automation technologies kept warehouses running even without significant human involvement.
Larger macroeconomic factors like rising domestic and international wages led companies to realize the ROI on one-time automation costs. Recent research suggests that industrial robot prices might drop by as much as 60% by 2025, making automation even more cost-effective.
The commoditization of enabling technologies like machine learning is also driving increased adoption of warehouse automation across key markets. LogisticsIQ suggests that the global warehouse automation market could double in size by 2025. Advanced robotic solutions could account for up to 41% of the $27 billion in annual spend. Warehouse automation can broadly be categorized into two components:
Process automation focuses on taking manual processes and varying workflows and adding automation components that reduce workload and manual hours.
In addition to process and physical automation, warehouse automation systems can also be categorized as basic augmentation (low level of automation like conveyor belts), software-based automation (process automation), RPA, and advanced automation (where a significant part of day-to-day operations have been automated).
You don’t need huge Amazon level fleets of autonomous mobile robots or massive automated conveyor belts to see the benefits of automation through machine learning based software. These tools can provide a ton of value your warehouse management system and pipelines you’re already using, or can become an entirely new piece of the backend puzzle.
Processing incoming invoices, vendor documents, and product labels requires manual effort and task understanding to take the information from these documents and input them correctly into backend systems. Most warehouses process a ton of these documents per day which requires workers to take time away from more important and time restrained tasks.
Automated document processing software leverages a number of machine learning and deep learning techniques to extract the valuable information from these documents and automatically update backend systems and databases. These models can be trained to handle a huge variety of document formats, sizes, and qualities of picture which allows for a higher accuracy in the real world. The pipeline setup can be as quick as snapping a picture of the document or label on a phone, and the software extracts the information and updates the backend systems in less than 3 seconds.
Here’s a look at the different technologies that go into a standard warehouse document automation system.
Deep learning based OCR algorithms allow us to extract the text from documents in a warehouse environment. These algorithms can read difficult text off box labels, handwritten characters, text from fuzzy or poorly taken images, and product images.
We leverage powerful pretrained algorithms like TrOCR as a backbone for our architecture then use in-domain fine-tuning to boost the accuracy of the product for your specific warehouse use case.
We use a deep learning based information extraction framework that allows us to extract text positioning, text size, key phrases, classification, and other parameters to understand the relationship between text fields in the document (‘QTY’, ‘Size’, ‘Item’ above) and the specific text that correlates.
We’re able to create reasoning through the relationship learned and support a number of downstream processes that require multiple variables or higher order variables.
Once the data has been extracted from these documents and mapped to any respective fields in the document, or some internal rules, we need to map to actual fields in our warehouse backend or database. Oftentimes the fields extracted for a document are either:
- Not exactly what we call them in our internal database for some business use reason. This happens when our internal system calls fields something different then what the document calls the same information
- The format that these fields are extracted in from the document does not fit our internal structure. This commonly happens when either the document or internal database combines or splits data into different fields.
We use a number of different NLP algorithms to postprocess this data into use case specific output.
Automated invoice matching allows you to automatically match text information for standardization across documents such as purchase orders, receipts, invoices, and the inspection report. This surprisingly manual task can be automated out using the same pipeline as above to extract key data like line items, product titles, quantity, price, and contract terms and validate it across all the required documents.
Product recognition software uses computer vision techniques like OCR and object detection to recognize areas where a product is located in an image or video stream, then uses embedded vector similarity to match the detected product area to a specific SKU or unique identifier in a database. This allows for the automation of a number of warehouse and logistics operations.
This architecture takes traditional object recognition a step further and provides a bunch of huge benefits for a warehouse environment:
1. The dynamic nature of comparing product area to a database allows you to evaluate products at a deeper level than you can with traditional object recognition. This lets you identify exact SKUs or sizes for the same overarching product.
2. The dynamic natural also allows you to support new products or box designs without needing to retrain the models at all. In traditional object recognition you would need to retrain the pipeline each time you want to add a new product or box design to the possible matches. With product recognition you just need to update the database images to add entirely new products or boxes.
3. You can get started with much less data and can scale to many more unique SKUs and IDs.
We deploy this as a cloud based rest API that allows you to use the pipeline on any website, phone app, or backend system with a camera. The pipeline returns product recognition results quickly and can provide alerts, confidence scores, and other management information.
Here’s an example from a real use case of how the benefits of this architecture can be used. Let’s say we’re a distributor of Coca Cola products that supports all Coca Cola products all year around. Both of the below boxes have different SKUs and must be recognized as different products.
To tell these two products apart with object detection we would need 100s of image examples with varying noise levels and multiple rounds of training. Any time the box changed to a seasonal design or throwback design we would have to train again.
To support this new box design for the 12 can pack we would have to collect images and retrain again. With the product recognition architecture we just have to add this image to the database and associate it with the unique ID of the original 12 can pack. The constant need for updating to new box designs, adding new SKUs, or other large changes to the product we are trying to recognize grows as we try to scale the softwares use.
Warehouses and Distribution Centers are always looking to better utilize their space based on current and future inventory. These businesses become overwhelmed due to seasonal changes, unforeseen late or early inventory, and poor space management. This results in poor efficiency in terms of stock going in and out, higher costs, and more manual labor required per inventory action. Planning and optimizing the required space is a difficult task to solve manually as the number of variables such as total warehouse space, forecasted inventory, demand optimization, and others make it difficult to model.
Warehouse space and slotting optimization algorithms work to setup the problem as a multi-variable optimization problem while using human input to determine the objectives and constraints of the slotting optimization. These algorithms have been shown to have huge ROI in warehouse operations such as 120-150% increase in current storage space, 37% productivity rate increase, and a huge reduction in operational costs (study).
The key gains come from a learned understanding of these concepts:
1. How much space a specific item(s) requires relative to how often it is accessed or moved.
2. The velocity of input to and output from the warehouse for a specific item relative to the above points based on historical and forecasted data. This data is often used in a simulation type environment to optimize available space for new products coming in and the shifts in volume in different seasons. One of the cool benefits is that your sales data acts as the only data needed for this step, which makes it easy to get off the ground.
3. What are the key objectives and constraints to this specific use case. This can be anything such as certain products having to be grouped together, number of laborers required to move a product, color of the product etc. We’re huge fans of focusing heavy on this step as with most ML/Ai problems a clear definition of the objectives makes it easier to reach a successful end.
Automated sortation systems have been around for years in manufacturing. These systems use RFID scanners and sensors to sort and divert products on a conveyor belt. Automated sortation systems are crucial to the entire fulfillment process, from picking to receiving and even shipping. These systems can integrate nicely with the product recognition architecture from above to sort products without RFID. Let’s look at the different types of sortation systems and applications:
data matching software uses natural language processing techniques to compare various product data fields to each other for similarity. These algorithms predict the similarity based on a learned relationship between these data points with a level of variance and noise. The cool part of how modern versions of these algorithms are designed allows for you to add or subtract fields on the fly and still reach a high level of accuracy. An example of this can be seen in our customer record matching system that allows fields like email, phone number, or username to be blank and still reach a production level system.
This software integrates perfectly with the document processing and product recognition use cases from above to further enhance the automation possibilities. We’ve fine-tuned the models to compare incoming product data from external sources such as vendors or sellers to an internal database of SKUs or unique IDs. This allows you to manage the incoming products and invoices from multiple sources to a single identifier while these external sources have different formats or field requirements. Here’s a UPC code matcher that can match extracted codes from documents or product labels with a huge noise variance level. The model even generates a reason for why these two UPC codes are match!
The first thing people usually associate with warehouse automation is improved ROI. However, in a modern warehouse, automation does much more than just help contain costs. The Amazon-ification of supply chain and logistics means that customers now have high expectations from brands when it comes to order fulfillment. Same-day deliveries, comprehensive tracking capabilities, and easy returns have become the price of entry for companies in the retail logistics business. The trend is gradually starting to take shape in B2B environments too.
Here’s why you should consider automation your warehouse operations:
Warehouses and distribution centers have an unique opportunity to develop their competitive advantage by leveraging machine learning and IoT for automation. However, identifying what sort of automation is needed and integrating it into your WMS is key. You need a solution that ensures that every step of your warehouse operation is optimized for automation.
Want to learn more about implementing a successful warehouse automation project? Book a free consultation with us to understand how our warehouse automation experience can fuel your growth.
Unlock the full potential of spaCy with this guide to building production-grade text classification pipelines for business data.
We compare 12 AI text summarization models through a series of tests to see how BART text summarization holds up against GPT-3, PEGASUS, and more.
Let’s take a look at what intent classification is in conversational ai and how you can build a GPT-3 intent classification model for conversational ai and chatbot pipelines.
Discover the capabilities of zero-shot object detection, which enables anyone to use a model out-of-the-box without any training and generate production-grade results.
What is facial expression recognition and what SOTA models are being used today in production
Get a simple TensorFlow facial recognition model up & running quickly with this tutorial aimed at using it in your personal spaces on smartphones & IoT devices.
Explore accurate classification algorithms using the latest innovations in deep learning, computer vision, and natural language processing.
Learn what human activity recognition means, how it works, and how it’s implemented in various industries using the latest advances in artificial intelligence.
What is the the SetFit architecture and how does it outperform GPT-3 and other few shot large language models
What is image classification and how we build production level TensorFlow image classification systems for recognizing various products on a retail shelf.
Explore the application of intelligent document processing (IDP) in different industries and dive in-depth on intelligent document pipelines.
How to build an image classification model in PyTorch with a real world use case. How you can perform product recognition with image classification
Let's build a custom CTA generator that you'll actually want to use for your website copy
We’re going to look at how we built a state of the art NLP pipeline for blended summarization and NER to process master service agreements (MDAs) that vary the outputs based on the input document and what is deemed important information.
Get a comprehensive overview of a purchase order vs. invoice, including when businesses use each, what information goes in them, and more.
Learn what Google Shopping categories are used for and how you can automate fitting products to this taxonomy using ai.
Automatically categorize your Shopify store products to the Shopify Product Taxonomy instantly with ai based PIM software
Dive deep into 3-way invoice matching, including how it works, eight benefits for your business, and the problems with doing it manually.
Smart farming using computer vision and deep learning provides the most promising path forward in the slow-moving industry of agriculture.
How we leveraged large language models to build a legal clause rewriting pipeline that generates stronger language and more clarity in legal clauses
Using ai for document information extraction to automate various parts of the loan process.
Apply AI to your favorite sport with this guide. Learn how automated ball tracking can change the game for coaches and players.
Categorize your ecommerce products to the 2021 google product taxonomy tree instantly with our Ai software
Surveying the current landscape of ecommerce automation and how you can use ai to automate huge chunks of your product management.
Classify your product data against an existing product category database or generate categories and tags in seconds using artificial intelligence
4 different NLP methods of summarizing longer input text into different methods such as extractive, abstractive, and blended summarization
iscover an invoice OCR tool that will revolutionize the way you handle invoices. There’s no human intervention needed & a dramatically lower per-invoice cost.
Instead of invoice matching taking upwards of a week, it could take mere seconds with the proper automation solution. Learn more here.
Manual and template-based invoicing are riddled with low accuracy and required human intervention. Learn how to systematically eliminate these issues with the right invoice data capture software.
A complete walkthrough guide on how to use visual search in ecommerce stores to create more sales and real examples of companies already using it.
Automating the extraction of data from invoices can reduce the stress of your accountants by finding inaccuracies, digitizing paper invoices, and more.
How you can use machine learning based data matching to compare data features in a scalable architecture for deduping, record merging, and operational efficiency
Learn how lifetime value or LTV prediction can improve your marketing strategies. Then, discover the best statistical & machine learning models for your predictions.
A deep understanding of how we use gpt-3 and other NLP processes to build flexible chatbot architectures that can handle negotiation, multiple conversation turns, and multiple sales tactics to increase conversions.
The popular HR company O.C. Tanner, which has been in business since 1927 and has over 1500 employees, was looking to research and design two GPT-3 software products to be used as internal tools with their clients. GPT-3 based products can be difficult to outline and design given the sheer lack of publicly available information around optimizing and improving these systems to a production level.
We’ll compare Tableau vs QlikView in terms of popularity, integrations, ease of use, performance, security, customization, and more.
With a context-aware recommender system, you can plan ways to recreate some of the contextual conditions that persuade them to buy more from you.
We’re going to walk through building a production level twitter sentiment analysis classifier using GPT-3 with the popular tweet dataset Sentiment140.
Find out how machine learning in medical imaging is transforming the healthcare world and making it more efficient with three use cases.
Discover ways that machine learning in health care informatics has become indispensable. Review the results of two case studies and consider two key challenges.
Accelerate your growth by pivoting key areas of your business to AI. Your business outcomes will be achieved quicker & you’ll see benefits you didn’t plan for.
We built a GPT-3 based software solution to automate raw data processing and data classification. Our model handles keyword extraction, named entity recognition, text classification | Case Study
We built a custom GPT-3 pipeline for key topic extraction for an asset management company that can be used across the financial domain | Case Study
How you can use GPT-3 to create higher order product categorization and product tagging from your ecommerce listings, and how you can create a powerful product taxonomy system with ai.
5 ways you can use product matching software in ecommerce to create real value that raises your sales metrics and improves your workflow operations.
Data mining and machine learning in cybersecurity enable businesses to ensure an acceptable level of data security 24/7 in highly dynamic IT environments. Learn how data security is getting increasingly automated.
Product recognition software has tremendous potential to improve your profits and slash your costs in your retail business. Find out just how useful it is.
Big data has evolved from hype to a crucial part of scaling your organization in every modern industry. Learn more about how big data is transforming organizations and providing business impacts.
Learn how natural language processing can benefit everybody involved in education from individual students and teachers to entire universities and mass testing agencies.
Here’s how automated data capture systems can benefit your business in some key ways and some real-life examples of what it looks like in practice.
Use these power ai and machine learning tools to create business intelligence in your marketing that pushes your business understanding and analytics past your competition.
We built a custom ML pipeline to automate information extraction and fine tuned it for the legal document domain.
In this practical guide, you'll get to know the principles, architectures, and technologies used for building a data lake implementation.
Find out how machine learning in biology is accelerating research and innovation in the areas of cancer treatment, medical devices, and more.
An enterprise data warehouse (EDW) is a repository of big data for an enterprise. It’s almost exclusive to business and houses a very specific type of data.
Dlib is a versatile and well-diffused facial recognition library, with perhaps an ideal balance of resource usage, accuracy and latency, suited for real-time face recognition in mobile app development. It's becoming a common and possibly even essential library in the facial recognition landscape, and, even in the face of more recent contenders, is a strong candidate for your computer vision and facial recognition or detection framework.
Learn how to utilize machine learning to get a higher customer retention rate with this step-by-step guide to a churn prediction model.
Machine learning algorithms are helping the oil and gas industry cut costs and improve efficiency. We'll show you how.
We’ll show you the difference between machine learning vs. data mining so you know how to implement them in your organization.
Here’s why you should use deep learning algorithms in your business, along with some real-world examples to help you see the potential.
Beam search is an algorithm used in many NLP and speech recognition models as a final decision making layer to choose the best output given target variables like maximum probability or next output character.
Best Place For was looking for an image recognition based software solution that could be used to detect and identify different food dishes, drinks, and menu items in images sourced from blogs and Instagram. The images would be pulled from restaurant locations on Instagram and different menu items would be identified in the images. This software solution has to be able to handle high and low quality images and still perform at the highest production level, while accounting for runtime as well as accuracy.
Deep learning recommendation system architectures make use of multiple simpler approaches in order to remediate the shortcomings of any single approach to extracting, transforming and vectorizing a large corpus of data into a useful recommendation for an end user.
Let's take a look at the architecture used to build neural collaborative filtering algorithms for recommendation systems
GPT-3 is one of the most versatile and transformative components that you can include in your framework, application or service. However, sensational headlines have obscured its wide range of capabilities since its launch. Let’s take a look at the ways that companies and researchers are achieving real-world results with GPT-3, and examine the untapped potential of this 'celebrity AI'.
How to get started with machine learning based dynamic pricing algorithms for price optimization and revenue management
Let's take a look at how you can use spaCy, a state of the art natural language processing tool, to build custom software tools for your business that increase ROI and give you data insights your competitors wish they had.
The landscape for AI in ecommerce has changed a lot recently. Some of the most popular products and approaches have been compromised or undermined in a very short time by a new global impetus for privacy reform, and by the way that the COVID-19 pandemic has transformed the nature of retail.
Extremely High ROI Computer Vision Applications Examples Across Different Industries
Building Data Capture Services To Collect High ROI Business Data With Machine Learning and AI
Software packages and Inventory Data tools that you definitely need for all automated warehouse solutions
Inventory automation with computer vision - how to use computer vision in online retail to automate backend inventory processes