kingjbbrooks

Deploying a Deep Learning Image Classification Model with NodeJS, Python, and Fastai

TL|DR: Use this to easily deploy a FastAI Python model using NodeJS.

You've processed your data and trained your model and now it's time to move it to the cloud.

If you've used a Python-based framework like fastai to build your model, there are several excellent solutions for deployment like Django or Starlette. But many web devs prefer to work in NodeJS, especially if your model is only part of a broader application.

My friend Navjot pointed out that NodeJS and Python could run together if we could send remote procedure calls from NodeJS to Python.

The Liang-Barsky algorithm is a cheap way to find the intersection points between a line segment and an axis-aligned rectangle. It's a simple algorithm, but the resources I was pointed to didn't have particularly good explanations, so I tried to write a better one.

Consider a rectangle defined by x_min ≤ x ≤ x_max and y_min ≤ y ≤ y_max, and a line segment from (x_0, y_0) to (x_0 + Δ_x, y_0 + Δ_y). We'll be assuming at least one of Δ_x and Δ_y is nonzero.

(I'm working with Flash, so I'll be using the convention that y increases as you go down.)

We want to distinguish between the following cases:

#	Name	Type 1	Type 2	Total	HP	Attack	Defense	Sp. Atk	Sp. Def	Speed	Generation	Legendary
1	Bulbasaur	Grass	Poison	318	45	49	49	65	65	45	1	False
2	Ivysaur	Grass	Poison	405	60	62	63	80	80	60	1	False
3	Venusaur	Grass	Poison	525	80	82	83	100	100	80	1	False
3	VenusaurMega Venusaur	Grass	Poison	625	80	100	123	122	120	80	1	False
4	Charmander	Fire		309	39	52	43	60	50	65	1	False
5	Charmeleon	Fire		405	58	64	58	80	65	80	1	False
6	Charizard	Fire	Flying	534	78	84	78	109	85	100	1	False
6	CharizardMega Charizard X	Fire	Dragon	634	78	130	111	130	85	100	1	False
6	CharizardMega Charizard Y	Fire	Flying	634	78	104	78	159	115	100	1	False

Option 1: Command-line download extension as zip and extract

extension_id=jifpbeccnghkjeaalbbjmodiffmgedin   # change this ID
curl -L -o "$extension_id.zip" "https://clients2.google.com/service/update2/crx?response=redirect&os=mac&arch=x86-64&nacl_arch=x86-64&prod=chromecrx&prodchannel=stable&prodversion=44.0.2403.130&x=id%3D$extension_id%26uc" 
unzip -d "$extension_id-source" "$extension_id.zip"

Thx to crxviewer for the magic download URL.

	/*
	* Chemical equation balancer
	*
	* Copyright (c) 2014 Project Nayuki
	* All rights reserved. Contact Nayuki for licensing.
	* http://www.nayuki.io/page/chemical-equation-balancer-javascript
	*/

	"use strict";

	var admobid={}
	if( /(android)/i.test(navigator.userAgent) ) {
	admobid = { // for Android
	banner: 'ca-app-pub-xxxxx/xxxxx',
	interstitial: 'ca-app-pub-xxxxxx/xxxxxx'
	}
	}
	else if(/(ipod\|iphone\|ipad)/i.test(navigator.userAgent)) {
	admobid = { // for iOS
	banner: 'ca-app-pub-xxxxxxx/xxxxxxx',

	AtomicNumber,Element,Symbol,AtomicMass,NumberofNeutrons,NumberofProtons,NumberofElectrons,Period,Group,Phase,Radioactive,Natural,Metal,Nonmetal,Metalloid,Type,AtomicRadius,Electronegativity,FirstIonization,Density,MeltingPoint,BoilingPoint,NumberOfIsotopes,Discoverer,Year,SpecificHeat,NumberofShells,NumberofValence
	1,Hydrogen,H,1.007,0,1,1,1,1,gas,,yes,,yes,,Nonmetal,0.79,2.2,13.5984,8.99E-05,14.175,20.28,3,Cavendish,1766,14.304,1,1
	2,Helium,He,4.002,2,2,2,1,18,gas,,yes,,yes,,Noble Gas,0.49,,24.5874,1.79E-04,,4.22,5,Janssen,1868,5.193,1,
	3,Lithium,Li,6.941,4,3,3,2,1,solid,,yes,yes,,,Alkali Metal,2.1,0.98,5.3917,5.34E-01,453.85,1615,5,Arfvedson,1817,3.582,2,1
	4,Beryllium,Be,9.012,5,4,4,2,2,solid,,yes,yes,,,Alkaline Earth Metal,1.4,1.57,9.3227,1.85E+00,1560.15,2742,6,Vaulquelin,1798,1.825,2,2
	5,Boron,B,10.811,6,5,5,2,13,solid,,yes,,,yes,Metalloid,1.2,2.04,8.298,2.34E+00,2573.15,4200,6,Gay-Lussac,1808,1.026,2,3
	6,Carbon,C,12.011,6,6,6,2,14,solid,,yes,,yes,,Nonmetal,0.91,2.55,11.2603,2.27E+00,3948.15,4300,7,Prehi

	My project was to allow users of the Godot game engine to create multiplayer games by connecting players via WebRTC using Godot's existing multiplayer APIs.

	Godot's core can be extended by adding modules - essentially plugins that are compiled with the engine. I created a WebRTC module that can be found here:
	https://github.com/BrandonMakin/godot/tree/webrtc_static_final/modules/webrtc

	That folder - modules/webrtc/ - represents my work for Google Summer of Code. SConstruct, config.py, are to help the build system compile and were created by my mentor Fabio Alessandrelli. register_types.h and register_types.cpp are there to expose the WebRTCPeer and NetworkedMultiplayerWebRTC classes to Godot's internal scripting and were created by Fabio Alessandrelli and modified by me, Brandon Makin. The rest of the code uses WebRTC's PeerConnectionInterface to implement Godot's multiplayer API. webrtc_peer.h and webrtc_peer.cpp create the WebRTCPeer class, a class that uses Godot's lower-level PacketPeer API.

	//
	// Matrix4+Extensions.swift
	// StARs
	//
	// Created by Konrad Feiler on 17.12.17.
	// Copyright © 2017 Konrad Feiler. All rights reserved.
	//

	import Foundation
	import SceneKit