--- # Dataset Card for Forlinx Embedded AI & Media Hub language: - en - zh license: apache-2.0 tags: - embedded-hardware - edge-ai - rockchip - rk3588 - nxp - imx95 - qwen - computer-vision - som - forlinx pretty_name: Forlinx Embedded Resource Hub --- # Forlinx Embedded: AI Models & Multimedia Resource Hub Welcome to the official **Forlinx Embedded** resource repository. This hub serves as a centralized ecosystem providing high-performance **Edge AI models**, **Hardware Development Kits**, and **Creative Marketing Assets** specifically tailored for our SoM (System on Module) and SBC (Single Board Computer) platforms. ## Resource Categories To accelerate your deployment and content creation, our resources are organized into three primary pillars: | Category | Key Deliverables | Target Platforms | | :--- | :--- | :--- | | **Edge AI Model Zoo** | NPU-optimized weights (Qwen, Llama, YOLO, MobileNet) | RK3588, i.MX95, RK3576, etc. | | **Development Resources** | Pre-built OS images, NPU toolchains (RKNN/eIQ), and SDKs | Forlinx SoM Boards | | **Creator Toolbox** | 4K product renders, application demos, and KOL B-roll footage | Marketers, Tech Bloggers, KOLs | --- ## 1. Hardware-Specific Ecosystems ### 🚀 Rockchip RK3588 Series * **NPU Capabilities:** 6 TOPS triple-core NPU support. * **Resources:** Optimized `.rknn` models, 8K hardware encoding/decoding samples, and multi-channel ISP configurations for high-end vision tasks. ### 🛡️ NXP i.MX95 Series * **NPU Capabilities:** Integrated eIQ Neutron Neural Processing Unit. * **Resources:** Industrial-grade safety demos, real-time control + AI fusion frameworks, and optimized models for machine vision and human-machine interface (HMI). --- ## 2. Optimized Model Zoo (Edge AI) We bridge the gap between heavy cloud models and resource-constrained edge devices: * **Large Language Models (LLM):** Qwen-1.5B/7B and Llama-3 (INT4/INT8 quantized) optimized for low-latency terminal interaction. * **Computer Vision:** YOLOv8, YOLOv10, and Segment Anything (SAM) tuned for real-time inference on Forlinx hardware. * **Multi-modal:** Speech-to-Text (Whisper) and Image Captioning models for interactive IoT devices. --- ## 3. Creative Media & KOL Support This section is dedicated to our marketing partners and content creators. * **High-Res Assets:** Transparent PNGs of Forlinx modules and 3D exploded view animations. * **Application Scenarios:** Raw video footage of smart medical, industrial automation, and smart city deployments for use in technical reviews. * **Brand Guidelines:** Official Forlinx logos, typography, and presentation templates for global consistency. --- ## How to Utilize These Resources ### For Developers 1. **Clone/Download:** Use `git lfs clone` or the Hugging Face CLI to fetch optimized model weights. 2. **Deployment:** Refer to the `deployment_guide.md` in each subdirectory for instructions on flashing images and running NPU inference. ### For Content Creators Navigate to the `/assets` or `/media` directory to find high-quality visuals. Please ensure the use of these materials aligns with the **Forlinx Brand Guidelines**. --- ## Technical Constraints & Risks * **Hardware Dependency:** AI models provided here (especially `.rknn` or `.nb` formats) are strictly compiled for specific NPU architectures. * **Environmental Factors:** Performance metrics (Inference speed/FPS) may vary based on thermal management and power supply configurations of the carrier board. ## About Forlinx Embedded Forlinx Embedded is a global leader in the design and manufacture of high-performance embedded solutions. As an official partner of **Rockchip, NXP, TI, and Allwinner**, we provide "Platform + Ecosystem" support to simplify the path from prototype to production. * **Global Website:** [www.forlinx.net](https://www.forlinx.net) * **Technical Support:** support@forlinx.com * **Sales Team:** sales@forlinx.com ## Licensing * **Code & Models:** Distributed under the **Apache 2.0** License unless otherwise specified. * **Media Assets:** Distributed under **CC BY-NC-ND 4.0** (Attribution-NonCommercial-NoDerivs).