When I read through below example in cpprefernce struct Abstract { virtual void f() = 0; // pure virtual }; // "Abstract" is abstract struct Concrete : Abstract { void f() override {} // non-pure virtual

# paho.c git clone https://github.com/eclipse/paho.mqtt.c cd paho.mqtt.c && mkdir build && cd build cmake \ -DPAHO_BUILD_STATIC=TRUE \ -DPAHO_WITH_SSL=FALSE \ -DPAHO_HIGH_PERFORMANCE=ON \ -DPAHO_ENABLE_TESTING=FALSE .. cmake --build . --config Release

WesleyCh3n - NTU-IoT-Server hackmd-github-sync-badge Reference Manually MQTT Mosquitto docker run -d --ip 172.17.0.3 \ --restart unless-stopped \ -p 1883:1883 \

Prepare pi-cross Pull the arch you need from pi-cross. In this example, aarch64, bullseye and g++ 10.2.0 are uesd for rpi 4 bullseye 64 bit os. docker pull wesleych3n/pi-cross:arm64-bullseye-10.2.0 Create helper script and grant it execution privilege docker run --rm wesleych3n/pi-cross:arm64-bullseye-10.2.0 > ./arm64-10.2.0 chmod +x ./arm64-10.2.0

What is singleton Singleton是整個類別(class)只會有一個實例(instance)存在，有利於系統控制整體的行為。 According to wiki: The singleton design pattern solves problems by allowing it to: Ensure that a class only has one instance Easily access the sole instance of a class Control its instantiation

Overview In train_*.py, it mainly contain the following parts: Dataset: tf.data.Dataset Model: tf.keras.models.Model Model Structure Training loop Loss

NTU-IoT-Cow NTU-IoT-Node NTU-IoT-Node NTU-IoT-Thermal NTU-IoT-Server NTU-IoT-Server: Overview NTU-IoT-Server: k3s

Reference Calculate anchor box with different ratio ./darknet detector calc_anchors \ data/obj.data \ -num_of_clusters 7 \ -width 416 -height 416 edit .cfg corresponding to the numbers of anchor boxes

Reference modify save_model.py change core/config.py python3 save_model_patch.py \ --weights ./face_model_default/yolov4-tiny-CowFace-anch-default_best.weights \ --output ./face_model_default/checkpoint/yolov4-tiny-416 \ --input_size 416 \ --model yolov4 \ --tiny

Installation Using python virtual environment (virtualenv). python3 -m virtualenv ansible # Create a virtualenv source ansible/bin/activate # Activate the virtual environment python3 -m pip3 install ansible Preparation ansible.cfg [defaults]

Step 1. Remove older version of CUDA/CuDNN Go to /usr/local/ see if there are any cuda-x.x folder. Then remove it with: sudo rm -rf /usr/local/cuda-x.x/ Step 2. Install CUDA Download from https://developer.nvidia.com/cuda-toolkit Choose the way with your environment,

Overview Baseline Model Training In baseline model training, we use Source domain dataset ( ). 1. Weight Initialization: Cross-Entropy Loss Edit exp/sample_experiment/baseline_softmax/params.py

Official Doc: cmake option reference Step (OpenCV Version: 4.5.0) sudo apt-get update sudo apt-get install libjpeg-dev libpng-dev \ libtiff-dev libavcodec-dev libavformat-dev \ libswscale-dev libv4l-dev libxvidcore-dev \ libx264-dev libatlas-base-dev

Reference In this example, we use the following configuration Server Wireguard ip: 10.112.0.1 Wireguard external port: 51871 Wireguard public key: rcflbneYW/3wVQy8H/jDi/oGlLgyrC4vmJvt4YJOVmw= Wireguard private key: yBo18fnFVjKrRS0dfH0DDehGrVBH1aDaZValIwdEW1I=

Pods 2 Pods will be created in each node. Stream pod: stream.t.yml Backup image pods: backup.t.yml stream.t.yml This pod run ntu-iot-node and pass the arguments it used. Mount Path:

This is NTU BME MS thesis project. The purpose of this project is using edge device (Rpi) to monitoring dairy cow feeding behavior. The code is focus on streaming also recognizing individual cow faces by edge computing (with Tensorflow Lite etc). This repo contains c++ and python version. The python version is deprecated, but still usable(in python branch). The main program uses c++. Usage $ ntu-node -h

Overview Introduction This is NTU BME MS thesis project backend setting. Include several services: Control: k3s: Control running services in nodes. Ansible: Update/Upgrade running program or configuration. WireGuard: VPN service for communication between server and nodes.