Application Catalog

The following is a catalog of prebuilt applications designed to accelerate AI development by providing ready-to-use models for well-known tasks. These applications include object detection, depth estimation, semantic segmentation, tracking, and pose estimation.

The catalog is divided into two sections:

  1. Developer Community Applications: Open-source or community-driven AI models for various domains.

  2. SiMa-Provided Applications: Optimized, hardware-accelerated models tailored for SiMa AI processors and multi-camera setups.

These prebuilt applications can significantly reduce development time and provide baseline models that can be fine-tuned for specific use cases.

Developer Community Applications

These applications are developed and contributed by the AI community, covering a wide range of deep learning tasks.

Developer Community Prebuilt Applications

Application Name

Description

SingleCameraIndoorsDepth

Optimized for depth estimation from an image includes Robotics Systems for navigation, object detection, and interaction.

LeatherAnomalyDetection

Optimized to identify defects and inconsistencies in leather products ensuring high-quality standards in leather processing, reducing manual inspection efforts and minimizing defects in the final product.

SafetyHelmetDetection

Designed for real-time monitoring and detection of safety hazards in various environments such as industrial sites, public spaces, and transportation systems.

YoloV7

Optimized for improved accuracy, speed, and efficiency this application is widely used for AI tasks such as autonomous driving, surveillance, industrial automation, and smart retail.

SiMa-Provided Applications

These applications are hardware-optimized for SiMa AI accelerators and multi-camera processing, ensuring high performance for edge AI use cases.

SiMa Prebuilt Applications

Application Name

Description

4_cameras

Multi-camera input processing for real-time applications

efficient_det

Efficient object detection model for constrained environments

open_pose

Human pose estimation model for activity recognition

people_tracking

Real-time tracking of individuals in video feeds

yolo_v5_ethernet

YOLOv5 deployment optimized for Ethernet-based processing

yolo_v5_pcie

YOLOv5 optimized for PCIe hardware acceleration

yolo_v7_4_cameras_ethernet

Multi-camera processing using YOLOv7 over Ethernet

yolo_v7_ethernet

YOLOv7 variant optimized for Ethernet communication

yolo_v7_pcie

YOLOv7 deployment optimized for PCIe acceleration