top of page

그라데이션 배경

Bos Semiconductors_logo_white

Eagle Software, Flexible enough to support any AI model, today and tomorrow

Eagle-N, Flexible enough to support any AI model, today and tomorrow

Eagle-N

Supports multiple data formats (INT8, FP8, BF16, FP16, Mx) and programmable RISC-V cores, enabling flexible support for current and future AI models.

Bos Semiconductors_Eagle-N

VLM and LLM

QWEN VLM

Qwen VLM is a large vision-language model that jointly understands images and text, enabling visual perception, reasoning, and multimodal interaction.

QWEN 2.5VL 7B_.png

블랙 노이즈 텍스처

QWEN 2.5VL 3B

LLaMA

LLaMA (Large Language Model Meta AI) is a family of efficient large language models designed for strong language understanding and reasoning with relatively low computational cost.

QWEN VL +
LLaMA

Qwen-VL + LLaMA combine vision-language understanding and large language reasoning to enable powerful multimodal perception and decision-making.

LLaMA 3.1 8B.png

블랙 노이즈 텍스처

LLaMA 3.2 1B

블랙 노이즈 텍스처

LLaMA 3.2 3B

QWEN 2_edited.jpg

QWEN 2.5-7B VLM + LLaMA

Object detection

SSR models enhance camera and vision perception by upscaling low-resolution inputs with minimal computational overhead.

YOLOv8s

YOLOv8s is a lightweight real-time object detection model that delivers fast and accurate performance for edge and embedded AI systems.

블랙 노이즈 텍스처

DeepLab V3+

DeepLab V3+ is a semantic segmentation model for accurate pixel-level scene understanding applications in vision.

블랙 노이즈 텍스처

PDL

PDL (Pathway Decomposition Learning) is a model designed for efficient and scalable perception processing tasks in vision.

블랙 노이즈 텍스처

FastOFT

FastOFT is a fast and efficient optical flow tracking model optimized for real-time motion estimation in dynamic vision systems.

ResNet50

ResNet50 is a 50-layer deep convolutional neural network that uses residual connections to enable stable and high-accuracy image recognition.

bottom of page