.svg)
Computer Vision Logistics Scanner
Springs team developed a tailor-made solution for the optimization of receiving processes on the large transshipping warehouse for the logistics provider.
Overview
- Client: Logistics Company
- Industry: Transportation, Logistics, Retail
- Location: Spain, EU
- Team: Python Developer, ML Engineer, React Developer, Node Developer, Project Manager, QA Engineer
- Timeline: 6 month
- Services: Discovery Phase, Web Development, ML Development, Quality Assurance, Project Management
- Tech Stack: Python, PyTorch, TensorFlow, Javascript, React, Node, MongoDB, GraphQL
- Problem: Optimize the handling of incoming packaged goods, with a keen emphasis on reducing both the time and human resources required.
- Product: Computer Vision ML Tool that efficiently recognizes and processes labels with integrated Admin Panel for managing all processes.
- Result: The creating of solution significantly helped company in operation cost savings and risk reduction.
Challenge
The client's primary objective revolved around optimizing the handling of incoming packaged goods, with a keen emphasis on reducing both the time and human resources required for these tasks before subsequent distribution. Recognizing the challenges posed by manual labor-intensive processes within their company, they were determined to find a solution that would introduce a level of automation through the power of machine learning.
Faced with a significant burden of manual labor and the need for greater efficiency, the client sought to transform their operations. Their top priority was to improve the processing of incoming packaged goods, with a particular focus on minimizing the time and manpower involved. To tackle these challenges head-on, they embarked on a journey to implement machine learning-driven automation, aiming to enhance their workflows and bolster overall productivity.
Product
The solution comprises two primary components: the Computer Vision ML Tool and the AI Web Application. At the heart of this solution lies a customized Warehouse Management System (WMS) module designed to streamline critical operations within the facility. This module harnesses the power of strategically positioned cameras, eliminating the fish-eye effect. Its core functionalities encompass two key tasks.
Label Recognition and Processing. The Computer Vision Tool efficiently recognizes and processes labels, reducing the need for manual intervention.
Generation and Printing of Unified Labels. It takes charge of generating and printing standardized labels throughout the facility, ensuring consistency and precision.
The Springs team has developed a user-friendly web application using JavaScript, specifically NodeJS and ReactJS. This application seamlessly integrates with a robust machine learning-based engine, driven by Python and PyTorch, dedicated to image and text recognition. Notably, an administrative panel has been included to offer users control over essential parameters, such as:
- Classes: The ability to manage and customize class definitions.
- Models: The capacity to configure and optimize machine learning models.
- Users: An interface to oversee and manage user accounts.
In summary, this two-part solution combines the power of Computer Vision with AI-driven web technology to enhance operational efficiency, reduce manual labor, and provide users with a versatile tool for image and text recognition, all while affording them the flexibility to tailor the system to their specific needs.
Result
The integration of ML and Web modules to Galvin achieved two essential goals: cost reduction and risk mitigation in the domains of goods receiving and forwarding. The successful implementation of this system hinged on two key steps: dataset acquisition and machine learning training. These steps were crucial to ensuring the high-precision recognition of marking text and efficient database matching.
- Cost Reduction and Risk Mitigation: by incorporating this module, the company managed to significantly cut down on expenses while simultaneously reducing potential risks associated with goods receiving and forwarding processes. This integration played a pivotal role in enhancing operational efficiency and minimizing the margin for errors, resulting in cost savings and a more secure operation.
- Dataset Gathering for Machine Learning: prior to system implementation, a critical step involved the collection of a comprehensive dataset. This dataset served as the foundation for machine learning, enabling the development of highly accurate text recognition capabilities. The dataset included various examples of marking text, creating a robust training environment for the machine learning model.
- High-Accuracy Text Recognition: with the dataset in place, the machine learning model was trained to achieve exceptional accuracy in recognizing marking text. This capability was instrumental in ensuring that the system could reliably identify and process labels and markings on goods.
- Efficient Database Matching: the recognition of marking text was closely followed by an efficient database search for a match. This step allowed for quick and precise retrieval of relevant information from the database, further enhancing the overall efficiency of the goods receiving and forwarding processes.
In summary, the integration of this module not only yielded cost savings and risk reduction but also underscored the importance of dataset preparation and machine learning training to achieve high-accuracy text recognition and efficient database matching.
Let's Discuss Your Project
If you have a project idea or require assistance with refining and estimating your concept, don't hesitate to schedule a meeting with our team. We are more than happy to provide our support!
Book a Call