---
title: AI inference with Yolo OBB model on EP-200Q
description: AI inference on EP-200Q's NPU with INT8 quantized Yolo OBB model.
---

[![Silex Technology](https://www.silextechnology.com/hubfs/raw_assets/public/silex/images/logo.svg "Silex Technology")](https://www.silextechnology.com/)

- [Connectivity Solutions](https://www.silextechnology.com/connectivity-solutions)
- [Services](https://www.silextechnology.com/connectivity-solutions/software-services) 
    - [Integration Services](https://www.silextechnology.com/de/software-service/integration-services)
    - [Customization Services](https://www.silextechnology.com/de/software-service/customization-services)
    - [Certification Services](https://www.silextechnology.com/software-service/global-certification-compliance)
- [Applications](https://www.silextechnology.com/application-expertise) 
    - [Medical ](https://www.silextechnology.com/medical)
    - [Transportation & Logistics](https://www.silextechnology.com/application-expertise/transportation-and-logistics)
    - [Factory Automation](https://www.silextechnology.com/application-expertise/factory-automation)
- [Resources](https://www.silextechnology.com/resources) 
    - [Resource Center](https://www.silextechnology.com/resources)
    - [Unwired Blog](https://www.silextechnology.com/unwired)
    - [Support Center](https://www.silextechnology.com/support)
    - [Security Center](https://www.silextechnology.com/security-center)
- [Company](https://www.silextechnology.com/company) 
    - [About Silex](https://www.silextechnology.com/company)
    - [Why Silex](https://www.silextechnology.com/why-silex)
    - [Technology Partnerships](https://www.silextechnology.com/why-silex/partner)
    - [News](https://www.silextechnology.com/unwired/topic/news)
    - [Careers](https://www.silextechnology.com/company/careers)

[ Where to buy ](https://www.silextechnology.com/where-to-purchase)

[ Contact sales ](https://www.silextechnology.com/contact-us)

This is a search field with an auto-suggest feature attached.

- There are no suggestions because the search field is empty.

![Embedded](https://www.silextechnology.com/hubfs/silex/image/Embedded.svg)

[ Embedded Wireless Modules ](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules)

- [Wi-Fi 7 (802.11be)](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules&categories=wi-fi-7-80211be)
- [Wi-Fi 6E (802.11ax)](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules&categories=wi-fi-6e-80211ax)
- [Wi-Fi 6 (802.11ax)](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules&categories=wi-fi-6-80211ax)
- [Wi-Fi 5 (802.11ac)](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules&categories=wi-fi-5-80211ac)

![Copy of Product icons home (40 x 40 px) (2)](https://www.silextechnology.com/hubfs/Copy%20of%20Product%20icons%20home%20(40%20x%2040%20px)%20(2).svg)

[ Platform & SOMs ](https://www.silextechnology.com/connectivity-solutions?services=platform--soms)

- [EP-200Q](https://www.silextechnology.com/connectivity-solutions/platform-and-soms/ep-200q)
- [EP-200N](https://www.silextechnology.com/connectivity-solutions-platform-and-som-ep-200n)

![Network System Solutions](https://www.silextechnology.com/hubfs/silex/image/Network%20System.svg)

[ Network System Solutions ](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions)

- [Ethernet to Wi-Fi Bridge](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=ethernet-to-wi-fi-bridge)
- [USB/Serial Connectivity](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=usb-serial-connectivity)
- [Wi-Fi HaLow Connectivity](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=wi-fi-halow-connectivity)
- [Wi-Fi Access Points](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=wi-fi-access-points)
- [Industrial Connectivity](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=industrial-connectivity)
- [Wireless Monitoring](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions&type=wireless-monitoring)

#### NEW: EP-200N AMC Edge SoM.

Powered by NXP® i.MX 95, delivering secure, safe and reliable connectivity for industrial and medical applications.

![EP-200N-01](https://www.silextechnology.com/hs-fs/hubfs/EP-200N-01.png?width=100&height=100&name=EP-200N-01.png)

[ Learn more ](https://www.silextechnology.com/connectivity-solutions-platform-and-som-ep-200n)

# Silex Knowledge Pool

# AI inference with Yolo OBB model on EP-200Q

Posted by Satoru Kumashiro, Nov 24, 2025 9:59:59 AM

#### OBB (Oriented Bounding Boxes) model for more accurate object detection

![](https://www.silextechnology.com/hs-fs/hubfs/image-png-Oct-12-2025-09-12-11-1162-PM.png?width=225&height=207&name=image-png-Oct-12-2025-09-12-11-1162-PM.png)

OBB model provides you with angle information of detected objects, resulting in more accurate detection of tilted objects on an image. One of the famous OBB model is Yolo OBB model, though there are other two-stage OBB detectors.

The example in this article uses Yolov8m-obb model. Before using the model on the EP-200Q, the model has to go through model conversion or quantization process.

Here is what we found with the OBB model on the EP-200Q.

#### INT8 quantization affects confidence scores, but minimally.

With the score more than 80%, all airplanes in the input picture are properly detected with the rotation information.  

![](https://www.silextechnology.com/hs-fs/hubfs/image-png-Oct-12-2025-10-59-26-6277-PM.png?width=463&height=463&name=image-png-Oct-12-2025-10-59-26-6277-PM.png) 

We also run the FP32 model, which is not quantized, on the CPU. All 29 airplanes are also detected as expected.

![](https://www.silextechnology.com/hs-fs/hubfs/image-png-Oct-12-2025-11-26-38-5025-PM.png?width=463&height=463&name=image-png-Oct-12-2025-11-26-38-5025-PM.png)

The average confidence scores of 29 airplanes in these two exercise are as below.

INT8 quantized model: 0.88

FP32 model: 0.885

As long as the confidence score is compared, there is the impact, but it is not significant.

However, if the application uses the confidence score as the threshold to recognize them as airplanes, careful review of the threshold value is necessary. When 0.9 is used for the threshold, only 1 airplane is recognized with INT8 model whereas 9 airplanes are recognized with FP32 model.

#### AI inference with INT8 quantized model on the EP-200Q NPU is superb.

We used a different image as below (100 planes in total) to check how fast AI inference process is.

The image is captured by a camera with 30fps frame rate. The captured image is fed into the AI inference process. Once the AI inference returns the value, the process proceeds with the next captured image. The program monitors how many images can be processed in one second.

![](https://www.silextechnology.com/hs-fs/hubfs/image-png-Oct-12-2025-11-39-54-2364-PM.png?width=522&height=276&name=image-png-Oct-12-2025-11-39-54-2364-PM.png)

![](https://www.silextechnology.com/hs-fs/hubfs/image-png-Oct-12-2025-11-44-03-6761-PM.png?width=520&height=301&name=image-png-Oct-12-2025-11-44-03-6761-PM.png)

|   | 640 x 640 input image size | 1024 x 1024 input image size |  |  |  |  |
| --- | --- | --- | --- | --- | --- | --- |
| model | FP32 | INT8 | INT8 | FP32 | INT8 | INT8 |
| Processor | CPU | CPU | NPU | CPU | CPU | NPU |
| FPS (Frame Per Second) | 0.79 | 3.04 | 29.8 | 0.31 | 1.19 | 28.96 |

 

The OBB model with either FP32 or INT8 can be run on CPU, but it takes time handle one image. The NPU can handle INT8 quantized model much faster. With the camera sensor with 30fps input, it can manage every single frame. When even slight improved confidence score is matter for the application, using FP32 model on the CPU could be still the solution, but please remind that running AI model in CPU can affect other tasks handed by the CPU.

 

#### Summary

It is demonstrated that the Yolov8n-obb model can be quantized and run on the EP-200Q NPU with decent accuracy and AI inference latency. When your application needs to detect several types of objects for the next processing on the CPU such as alerting, logging, controlling something in real time matter, the EP-200Q could be the solution for your system.

 Share

- [Tweet](https://twitter.com/share)

Topics

- [EP-200Q](https://www.silextechnology.com/unwired/topic/ep-200q),
- [Edge AI](https://www.silextechnology.com/unwired/topic/edge-ai),
- [AI App Development](https://www.silextechnology.com/unwired/topic/ai-app-development),
- [Use Cases](https://www.silextechnology.com/unwired/topic/use-cases)

### Related Posts

[Posted by Satoru Kumashiro, November 24, 2025 How does Silex support developers to setup the development environment? What is available from Qualcomm? Qualcomm has prepared extensive amount of documents for developers to develop firmware for their Dragonwing solutions...Keep Reading ](https://www.silextechnology.com/platform-and-som-knowledge-pool/how-does-silex-support-developers-to-setup-the-development-environment)

[Posted by Satoru Kumashiro, November 24, 2025 EP-200Q: Built for Longevity, Engineered for Performance The EP-200Q is designed for product longevity At Silex, we understand that product longevity is key to the success of any embedded system, especially ...Keep Reading ](https://www.silextechnology.com/platform-and-som-knowledge-pool/ep-200q-built-for-longevity-engineered-for-performance)

#### Subscribe to Email Updates

### Posts by Topic

- [EP-200Q (13)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/ep-200q)
- [Edge AI (10)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/edge-ai)
- [Knowledges (8)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/knowledges)
- [AI App Development (7)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/ai-app-development)
- [Features (6)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/features)
- [Hardware (6)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/hardware)
- [Use Cases (6)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/use-cases)
- [Product Ovewview (4)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/product-ovewview)
- [Software (4)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/software)
- [Feature Description (2)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/feature-description)
- [Yocto Development (2)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/yocto-development)
- [Connectivity (1)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/connectivity)
- [Evaluation Kit (1)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/evaluation-kit)
- [Silex SDK (1)](https://www.silextechnology.com/platform-and-som-knowledge-pool/tag/silex-sdk)

[See all](https://www.silextechnology.com/platform-and-som-knowledge-pool/ai-inference-with-yolo-obb-model-on-ep-200q#)

[![Silex Technology](https://www.silextechnology.com/hubfs/raw_assets/public/silex/images/logo.svg "Silex Technology")](https://www.silextechnology.com/)

### Subscribe to our blog

- [**](https://x.com/silextechnology)
- [**](https://www.linkedin.com/company/silex-technology-america/)
- [**](https://www.youtube.com/user/SilexTechnology)

##### Solutions

- [Embedded Wireless Modules](https://www.silextechnology.com/connectivity-solutions?services=embedded-wireless-modules)
- [Platform & SOMs](https://www.silextechnology.com/connectivity-solutions?services=platform--soms)
- [Network System Solutions](https://www.silextechnology.com/connectivity-solutions?services=network-systems-solutions)
- [Services](https://www.silextechnology.com/connectivity-solutions/software-services)

##### Applications

- [Medical Device Connectivity](https://www.silextechnology.com/application-expertise/medical-devices-learn-more)
- [Transportation & Logistics](https://www.silextechnology.com/de/application-expertise/transportation-and-logistics)
- [Factory Automation](https://www.silextechnology.com/application-expertise/factory-automation)

##### Resources

- [Resource Center](https://www.silextechnology.com/resources)
- [Unwired Blog](https://www.silextechnology.com/unwired)
- [Support Center](https://www.silextechnology.com/support)

##### Company

- [About Silex](https://www.silextechnology.com/company)
- [Why Silex](https://www.silextechnology.com/why-silex)
- [Partnerships](https://www.silextechnology.com/why-silex/partner)
- [News](https://www.silextechnology.com/unwired/topic/news)
- [Careers](https://www.silextechnology.com/company/careers)

##### Contact

- [U.S. Sales](mailto:sales@silexamerica.com)
- [Europe Sales](mailto:sales@silexeurope.com)
- [Customer Support](https://www.silextechnology.com/support/contact-silex-support)

© 2026 Silex Technology America, Inc. [Privacy Policy](https://www.silextechnology.com/contact-us/privacy-policy) [Terms of Use](https://www.silextechnology.com/contact-us/terms-of-use)

![](https://px.ads.linkedin.com/collect/?pid=8007468&fmt=gif)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Satoru Kumashiro",
    "url" : "https://www.silextechnology.com/platform-and-som-knowledge-pool/author/satoru-kumashiro-1"
  },
  "dateModified" : "2026-02-06T21:53:50.011Z",
  "datePublished" : "2025-11-24T17:59:59.000Z",
  "headline" : "AI inference with Yolo OBB model on EP-200Q",
  "mainEntityOfPage" : {
    "@id" : "https://www.silextechnology.com/platform-and-som-knowledge-pool/ai-inference-with-yolo-obb-model-on-ep-200q",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://www.silextechnology.com/hubfs/Marketing%20and%20Branding/SilexLogo.png"
    },
    "name" : "Silex"
  }
}
```