感知识别算法的系统环境部署流程

1. Conda安装

# 安装Miniconda，官网目录（https://repo.anaconda.com/miniconda/）
# 本项目下载linux对应的python3.10
wget https://repo.anaconda.com/miniconda/Miniconda3-py310_25.1.1-2-Linux-x86_64.sh
chmod +x Miniconda3-py310_25.1.1-2-Linux-x86_64.sh
sh Miniconda3-py310_25.1.1-2-Linux-x86_64.sh

# 更换Conda源
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --show-sources
# 更换pip源
pip config set global.index-url https://mirrors.aliyun.com/pypi/simple/
pip config list

2. GPU环境配置

nano ~/.bashrc
export PATH="/home/abner/miniconda3/bin:$PATH"

# Nividia驱动安装
sudo ubuntu-drivers devices # 查看系统支持及推荐版本
sudo apt purge nvidia-* && sudo apt autoremove
sudo apt install nvidia-driver-470

# CUDA 安装
nvidia-smi # 查看已有版本
# 安装指定版本（11.8）的CUDA版本：https://developer.nvidia.com/cuda-toolkit-archive
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-ubuntu1804.pin
sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda-repo-ubuntu1804-11-8-local_11.8.0-520.61.05-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1804-11-8-local_11.8.0-520.61.05-1_amd64.deb
sudo cp /var/cuda-repo-ubuntu1804-11-8-local/cuda-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install cuda

# 创建新环境
conda env list #查看已有虚拟环境清单
conda create --name pytorch22_py310 python=3.10
conda activate pytorch22_py310
# pytorch 安装
# 确认适配的pytorch版本： https://pytorch.org/get-started/locally/
pip3 install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cu118
conda deactivate
conda create --name robot_pytorch22_py310 --clone pytorch22_py310

3. 项目环境配置

# 基础环境
cd ~/model
sudo apt install libjpeg-dev zlib1g-dev
git clone https://github.com/pytorch/vision torchvision
cd torchvision
git checkout v0.16.2
pip install pillow==10.4.0
python3 setup.py install

# 非正常热源识别项目 （efficientvit 文件夹）
cd perceptionalg/model/efficientvit
conda activate robot_pytorch22_py310
pip install -e ./
pip install -r ../pysenxor/requirements_windows.txt
pip install supervision
pip install numpy==1.26.4

# 物品识别（YoloV10目录）
cd perceptionalg/model/yolov10
pip install -r requirements.txt

4. 程序运行

# 非正常热源识别(需要插入热成像摄像头)
cd perceptionalg/model/efficientvit
sudo chmod 777 /dev/ttyACM0
python scripts/grounding_dino-efficientvit_sam-thermal.py --video assets/light.mp4

# 物品识别（YoloV10目录）
cd perceptionalg/model/yolov10
python track.py --video assets/cat.mp4

5. 原理

1. 非正常热源识别
   对于采集的相机视频或视频文件，算法识别出灯、微波炉等家庭热源物体，然后在热感数据视频中，采集对应区域温度均值作为被识别物体的温度。

2. 物品识别

因为文件较大删除的文件清单

或者，可以使用LFS增加文件
查询前10个最大文件

find ./* -type f -exec du -h {} + | sort -hr | head -n 10
944M	./data/iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pt
660M	./data/IDEA-Research/grounding-dino-tiny/pytorch_model.bin
658M	./data/IDEA-Research/grounding-dino-tiny/model.safetensors
205M	./data/iic/speech_eres2netv2w24s4ep4_sv_zh-cn_16k-common/pretrained_eres2netv2w24s4ep4.ckpt
167M	./data/insightface/models/buffalo_l/w600k_r50.onnx
163M	./data/jetson/torch-2.1.0a0+41361538.nv23.06-cp38-cp38-linux_aarch64.whl
137M	./data/insightface/models/buffalo_l/1k3d68.onnx
133M	./data/mit-han-lab/efficientvit-sam/efficientvit_sam_l0.pt
113M	./data/ultralytics/best.pt