Page 563 - ISC PROCEEDINGS 21.4

spending, businesses had to accept smaller orders, rush the production, and shorten
delivery times, reducing profit margins and directly impacting key product groups.
Traditional markets faced significant pressure. This context compelled businesses to seek
new markets, accept scattered orders, and continuously adjust production plans.
Companies also needed to meet green standards and traceability requirements to
maintain credibility with international customers. Notably, in the context of geopolitical
fragmentation and a host of new tariff barriers, from CBAM (Border Carbon Adjustment
Mechanism) to stricter rules of origin, global supply chains are becoming increasingly
vulnerable. Therefore, brands are forced to diversify production locations to reduce risk.
Furthermore, the Vietnamese textile and garment industry mainly participates in the
production stage and has not yet developed strongly in high-value-added stages such as
design, branding, or distribution. On the other hand, Vietnam has lost its labour cost
advantage compared to many other exporting countries. Consequently, basic large-
volume orders with low processing costs are shifting to countries with cheaper labor cost.
With a focus on sustainability and the digital economy, the industry is strongly
shifting towards FOB (Free On Board) and ODM (Original Design Manufacturer) models in
smart factories. A smart factory integrates a technology platform with a system of
machinery and equipment connected to the Internet of Things (IoT). This setup allows
data to be aggregated and analyzed using artificial intelligence (AI) applications. In
traditional garment factories and workshops, the production process of separate
departments performing different tasks is divided among various departments, each
performing separate tasks. In contrast, in smart factories, these departments are linked
and connected together using technology. IoT brings unification to all stages and
departments in the production process, both within the factory and across the entire
enterprise. All departments, such as offices, production workers, quality control,
warehousing, and shipping, are interconnected to form a unified system. Specifically,
some businesses have already implemented sensor and IoT systems:
+ Installing sensors on machinery and production lines to collect real-time data
(temperature, pressure, operating speed, process errors, etc.) as well as images/videos
from surveillance cameras. This data serves as the "fuel" for AI models.
+ Building a data lake/data warehouse: Gathering and storing historical data from
machinery, quality control systems, and productivity data – to support AI model training
and data analysis.
+ Using computer vision solutions to analyze product images during production,
using algorithms for error detection, print pattern analysis and fabric uniformity testing
which help detect errors early and reduce waste.
+ Equipment forecasting and maintenance: Applying machine learning models to
analyze operational data from sensors allows for the prediction of when maintenance is
needed, preventing unexpected breakdowns. This solution reduces downtime and
optimizes maintenance costs.
+ Optimizing the production line: Deploying AI Agents to analyze production data
and predict bottlenecks. By optimizing production schedules, adjusting machine speeds,
and allocating resources, the system will help improve the overall efficiency of the
production line … Thanks to these measures, managers will easily manage ongoing factory
operations, making appropriate and quick adjustments when errors or problems are
detected in the processes. The application of smart garment factories will have positive
impacts on Vietnamese textile and garment businesses, such as the ability to remotely

562