floraachy
/
SpireCV
forked from floratest1/SpireCV
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
128 Commits 10 Branches 2 Tags 762 KiB
C++ 89.9%
Shell 5.2%
CMake 1.6%
Cuda 1.6%
C 1.1%
Other 0.6%
Go to file
Daniel 894717a47e Modify the Bytetrack part of the Json. 2024-05-25 18:37:23 +08:00
algorithm add a new mot Bytetrack. 2024-05-25 18:29:11 +08:00
gimbal_ctrl Fixed a bug caused by abnormal timestamps in the ROS environment 2024-04-11 07:52:39 +00:00
include add a new mot Bytetrack. 2024-05-25 18:29:11 +08:00
params Modify the Bytetrack part of the Json. 2024-05-25 18:37:23 +08:00
samples eval sort on mot17 2024-04-22 00:03:55 +08:00
scripts add opencl hw decoding in gst and intel acceeleration's opencv that reduce double the CPU usage rate in a sot task. 2024-05-25 17:19:22 +08:00
utils finish model manager v1 2024-01-12 21:58:08 +08:00
video_io add opencl hw decoding in gst and intel acceeleration's opencv that reduce double the CPU usage rate in a sot task. 2024-05-25 17:19:22 +08:00
.gitignore add MJPG 2024-01-01 15:03:31 +08:00
CMakeLists.txt add a new mot Bytetrack. 2024-05-25 18:29:11 +08:00
LICENSE fst commit 2023-06-16 10:42:02 +08:00
README.en.md fix 2024-01-01 11:18:37 +08:00
README.md fix 2024-01-01 11:18:37 +08:00
build_on_jetson.sh fst commit 2023-06-16 10:42:02 +08:00
build_on_x86_cuda.sh fst commit 2023-06-16 10:42:02 +08:00
build_on_x86_intel.sh add x86-intel in cmake & testcase-fps 2023-07-11 15:11:35 +08:00

README.en.md

SpireCV logo

SpireCV Intelligent perception algorithm library

Overview

SpireCV is an real-time edge perception SDK built for intelligent unmanned systems, with main functions including camera/pod control, video saving and push streaming, target detection and tracking, and edge data management. It aims to provide mobile robot developers with high-performance, high-reliability, simple interface and feature-rich visual perception capabilities.

Quick start

  • Installation and use: SpireCV user manual、SpireCV developer kit guide

    • Basic knowledge of C++ language and CMake compilation tool is required
    • Need to master the foundation of OpenCV vision library and understand the computational libraries such as CUDA, OpenVINO, RKNN and CANN
    • Need to understand the basic concepts and basic operation of ROS

  • Q&A and communication:

    • Q&A forum (official regular Q&A, recommended): Amovlab Community - SpireCV Q&A Zone
    • Add WeChat yinyue199506 (note: SpireCV) into the SpireCV intelligent perception algorithm library exchange group
    • Search and subscribe "Amovlab" on YouTube.com, we will update the video from time to time

Framework

The main framework is shown in the figure:

Current support:

  • Functional level：
    • Video algorithm module (providing perceptual algorithms with unified interfaces, efficient performance and diverse functions)
    • Video input, save and push stream module (to provide stable, cross-platform video reading and writing capabilities)
    • Camera and pod control module (for the typical hardware ecology to open the interface, easy to use)
    • Sensing information interaction module (providing UDP communication protocol)
    • ROS interface
  • Platform level：
    • X86 + Nvidia GPUs (10 series, 20 series, and 30 series graphics cards recommended)
    • Jetson (AGX Orin/Xavier、Orin NX/Nano、Xavier NX)
    • Intel CPU
    • HUAWEI Ascend (coming soon)
    • Rockchip (coming soon)

Demos

  • QR code detection

  • Landing mark detection

  • Ellipse detection

  • Target click tracking (including target detection and tracking)

  • Low latency push streaming

Copyright statement

  • SpireCV is protected under the Apache License 2.0.
  • SpireCV is for personal use only, please do NOT use it for commercial purposes.
  • If this project is used for profit, Amovlab will pursue infringement.