Utilizing maker technology to help understand
environments around us
In addition to my work in research, I also enjoy spending time working with Maker Technology to develop tools to help me understand the environment around us.
These projects are not only interesting to me but also help me improve as an engineer as well. They lie in the intersection between my interest and my work such that I can also apply techniques I learn from these projects to my research work, and vice versa.
A real-time bird classification with deep learning and Jetson Nano
An environmental sensor network for home monitoring
Custom-made photography gears for one-of-a-kind photos
A real-time bird classification with deep learning and Jetson Nano
During 2020, I spent a lot of time at home and started to notice a lot of birds coming to my garden, so I was curious to learn their behaviors at my bird feeder. As a result, I develop BirdCam to perform real-time classification of birds and other faunas in my garden using Jetson Nano.
► Utilizes traditional image processing to gather training images and only crop regions of interest.
► Uses modern machine learning with convolutional neural network to perform classification.
► A MobileNet V2 on Tensorflow framework is optimized with TensorRT for faster performance on Jetson Nano.
► Achieve a real-time (>20 fps) classification of 7 classes.
► Real-time classification allows instantaneous execution of processes, such as using a water gun to deter rats from the feeder (see video below).
Detailed description of the project and source code can be found on Github linked below.
An environmental sensor network for home monitoring
Interesting environmental phenomena are happening all around us. Wouldn't it be interesting to know how the environments affect us?
Examples:
► How much indoor temperature and humidity changes during a Southern California heatwave?
► How can we decrease the indoor level of CO2?
To answer these questions, I developed an environmental sensor network, "Evniron," to continuously collect interesting environmental parameters around my home.
○ Each unit is controlled by an ESP32 microcontroller, which collects data from sensors and directly send them to store in the oceanleaf server.
○ The server processes the data and displays them on the website, showing both real-time and historical data.
See an example unit below.
Custom-made photography gears for one-of-a-kind photos
As an engineer and a photographer, I am passionate about using new technology to help improve quality of photos. I developed my own tools to help me capture unique pictures.
Follow my latest photography work on my Instagram and see selected photos on the photography page.
Created with
Mobirise