flashstudio 0.2.1

Interactive Training & Inference UI for FlashDet — runs on Google Colab

FlashStudio

Interactive Training & Inference UI for FlashDet

---

Install

pip install flashstudio
pip install git+https://github.com/FlashVision/FlashDet.git

Quick Start

# Launch locally
flashstudio

# Or with Python
from flashstudio import launch
launch()

Google Colab

!pip install flashstudio
!pip install git+https://github.com/FlashVision/FlashDet.git

from flashstudio import launch
launch(ngrok_token="YOUR_TOKEN")

Get a free ngrok token at: https://dashboard.ngrok.com/get-started/your-authtoken

Features

• Training — Launch and monitor FlashDet training with live loss curves
• Inference — Run detection on images/video with 17 built-in solutions
• Model Config — FlashDet (Pico → X) + YOLOv8/v9/v10/v11/YOLOX
• Data — Upload datasets or download from FlashDet registry
• Export — ONNX export with FP16 support
• Dashboard — Project overview with recent runs and metrics

Supported Models

Model	Params	Use Case
FlashDet-Pico	~298K	Edge / MCU
FlashDet-Nano	~790K	Embedded / IoT
FlashDet-Small	~1.8M	General purpose
FlashDet-Medium	~3.6M	High accuracy
FlashDet-Large	~5.8M	High accuracy
FlashDet-X	~9.0M	Server / Max accuracy

Requirements

• Python >= 3.10
• FlashDet (for training/inference)
• ngrok account (free, for Colab)
• GPU recommended for training

Links

• GitHub: https://github.com/FlashVision/FlashStudio
• FlashDet: https://github.com/FlashVision/FlashDet
• Organization: https://flashvision.github.io

License

Apache-2.0