vercel-labs
Featured
scriptc
scriptc compiles ordinary TypeScript into small, fast native executables without needing Node.js, V8, or any JavaScript runtime in the binary.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
It demonstrates a practical edge-AI application for self‑tracking distraction, with a custom data collection pipeline and deployment on embedded hardware.
Who it is for
- Developers interested in embedded computer vision on Jetson
- Individuals seeking automated productivity monitoring
- Students learning end‑to‑end model training and deployment
- Hobbyists building personal AI assistants
Use cases
- Real‑time self‑monitoring of work vs. distractions
- Educational reference for training a custom classifier on Jetson
- Prototyping a low‑cost attention‑tracking system
- Testing data collection and model iteration workflow
Strengths
- End‑to‑end pipeline from data capture to inference
- Runs on edge device (Jetson Orin) for privacy and low latency
- Lightweight ResNet‑18 model suitable for real‑time use
- Clearly documented setup steps for replicating the environment
Considerations
- Validation set too similar to training set, risking overfitting
- Only three activity classes limit comprehensive productivity measurement
- Requires specific NVIDIA Jetson hardware and jetson‑inference setup
README quick start
productivity-checker
This project uses a model trained in jetson-inference to measure productivity levels when I'm working. Since I get distracted very easily when I should be doing work, I thought this project would be both a great learning experience and a way to solve a real world problem that I encounter on a daily basis.
Overview
This project uses a resnet-18 model which was trained on 3 classes, sleeping, looking at phone, and working. The dataset used a custom script located in script.py that takes multiple pictures on a camera feed, and saves them into a specified directory. The model is trained on this data and productivity levels are measured based on the confidence of the model's inference.
Classification Model
Model Summary
Model Metrics
Set up
- SSH into orin
- Install necessary packages
sudo apt-get update
sudo apt-get install -y git cmake build-essential libpython3-dev python3-pip python3-numpy - Clone jetson-inference
git clone --recursive --depth=1 https://github.com/dusty-nv/jetson-inference - Update submodules
git submodule update --init --recursive - Remove npymath because of a compatibility issue
sed -i 's/ npymath//g' python/bindings/CMakeLists.txt
sed -i 's/ npymath//g' utils/python/bindings/CMakeLists.txt - Install cmake
mkdir build
cd build
cmake ../
Clone the repository:
git clone git@github.com:kezhang29/productivity-checkers.git
Run the project:
python3 model.py
Next Steps
- The validation set resembles the training set too much, causing the steep jump in validation accuracy shown in tensorboard. Retraining with a more diverse dataset would improve model results.
- Add more labels
Related repositories
Similar projects matched by category, topics, and programming language.
jamesob
Featured
local-llm
A comprehensive guide for building and configuring a high-end local machine to run state-of-the-art LLMs, with detailed hardware choices, BIOS tuning, and Docker-based model serving.
mereyabdenbekuly-ctrl
Featured
clodex-ide
Clodex is an open-source, local-first agentic IDE that combines persistent AI tasks, code, terminal, browser, Git, models, memory, and governed execution in one Electron workspace, currently in technical preview.