nitesh-0/times-net

How nitesh-0/times-net works

This repository implements a RetNet video action-recognition demo: a user uploads a video, selects a dataset and inference mode, and views predictions and performance metrics. The Streamlit frontend sends parallel requests or consumes a streaming response from a Modal-hosted FastAPI backend. Backend excerpts show video extraction and preprocessing, model-related components, and parallel/streaming inference helpers, but omit much of the routing and inference implementation; dashed, uncited edges mark those unverified connections.

Main entry points: modal_app.py Parallel endpoint · app.py Parallel client

Components 13

Demo frontend

Mode and dataset frontend/app.py
Video upload frontend/app.py
Parallel client frontend/app.py
Streaming client frontend/app.py
Predictions and metrics frontend/app.py

Backend API

Parallel endpoint backend/modal_app.py
Streaming endpoint backend/modal_app.py
Frame extraction backend/modal_app.py
Frame preprocessing backend/modal_app.py

Inference runtime

Model components backend/modal_app.py
Parallel inference backend/modal_app.py
Streaming inference backend/modal_app.py

Other components

Video user

Connections 17

Video user → Video upload: uploads video frontend/app.py
Video user → Mode and dataset: selects mode frontend/app.py
Video upload → Parallel client: submits video frontend/app.py
Mode and dataset → Parallel client: selects request frontend/app.py
Parallel client → Parallel endpoint: posts video frontend/app.py
Video upload → Streaming client: submits video frontend/app.py
Mode and dataset → Streaming client: selects request frontend/app.py
Streaming client → Streaming endpoint: streams response frontend/app.py
Parallel endpoint → Frame extraction: extracts frames backend/modal_app.py
Parallel endpoint → Frame preprocessing: preprocesses frames backend/modal_app.py
Streaming endpoint → Frame extraction: extracts frames backend/modal_app.py
Parallel endpoint → Parallel inference: runs inference backend/modal_app.py
Streaming endpoint → Streaming inference: runs inference backend/modal_app.py
Parallel inference → Model components: uses models backend/modal_app.py
Streaming inference → Model components: uses models backend/modal_app.py
Parallel client → Predictions and metrics: renders results frontend/app.py
Streaming client → Predictions and metrics: renders updates frontend/app.py

Drawn by GitDiagram on November 10, 2026 from the repository's file tree, README and source files. Read as Markdown ·