Everything

A feed of all the things :)

Tab Clearning (December 4 2024)

misc

Clearing out my open tabs

AI Outputs as Junk Food

essays

My first taste of American food was a bacon-and-egg roll in the airport. Immediate impression: “woah this is sweet?!?”. The bread was sweet, the bacon maple-coated… Tasty…

Making Giant Bubbles

mini-hw-projects

It turns out it’s pretty easy to make GIANT bubbles! This feels like the kind of knowledge that should be more widespread.

Nerdsnipe: Map Folding

mini-projects

Nerd-sniped by a video from the channel ‘Physics for the Birds’, I looked into what it would take to add an extra item to this sequence of integers: https://oeis.org/A001418 …

3 Modes of Making with AI

wip

essays

In this post I’ll share a mental model that I’ve found useful recently in thinking through a number of topics related to AI and creativity. I call it the “3 Modes of Making…

Deploy Your Artifacts!

I’m a big fan of the idea of Sharing Scrappy Fiddles.

Facewave: Controlling Midi Expression with your webcam

mini-projects

One gripe I’ve had with digital instruments is a lack of expression. Coming from a wind instrument background, I don’t like having no control over a note once it starts. So…

Mini Synth and Audio Visualization

mini-projects

I’ve been enjoying creating lots of little projects with Claude artifacts and other tools (post on that soon). Two especially fun ones were both audio related: a mini synth…

Tab Clearning (October 7 2024)

misc

New Reka Flash model - I’m excited to see them ticking along, doing what looks like good work making evals then improving on them (e.g. longer context, instruction…

Tab Clearning (September 23 2024)

misc

https://openai.com/global-affairs/openai-academy/: “Investing in the development of local AI talent can fuel economic growth and innovation across sectors like healthcare…

Doodling In Hardware

mini-hw-projects

I wanted to do something physical. Maybe a motion system that’s not just another cartesian X/Y thing? Something I can test out Kalman filters and control systems on?…

TIL: Kalman Filters

tils

This is a notebook I used to explore the topic of Kalman Filters, while attending a talk on the same. You can see I leveraged AI heavily - the presentation here is roughly…

Min P Sampling: Balancing Creativity and Coherence (paper explanation + code)

Video

exploring min-p sampling

Single-Pixel Camera (Hardware)

mini-hw-projects

Taking photos one pixel at a time is a funny idea. It’s been done a few tmes (e.g. https://hackaday.com/2024/07/23/photoresistor-based-single-pixel-camera/, love the…

Building Silly Projects with FastHTML

Blog

We’ve been working on FastHTML, a web app framework that makes it easy to build web apps with Python. In the process, I personally have learnt a lot about how the web works…

Computational Linguistics

Blog

LLMs are cool - we can make them do all sorts of tricks. But beyond the generative applications there’s a whole bunch of interesting linguistic analysis we can do…

What You Should Read (AI Edition)

Blog

There have been hundreds of thousands of films made. But if you reveal that you haven’t seen Star Wars, you’re regarded with a mixture of sympathy and scorn. In this…

Mixed-Modal Early-Fusion Foundation Models: Paper run-throughs for ‘Chameleon’ and ‘MoMa’

Video

Text-only LLMs are great, and we’ve seen people pasting on some image support here and there, but the future it seems is multi-modal. What does it take to train models from scratch that take in both images and text (and more)? In this video we look at two key papers from FAIR at Meta, introducing their Chameleon approach and making it more efficient with mixture of experts.

Robot Arm Shenanigans (Draft Post)

mini-hw-projects

I built a fun little robot arm based on designs from here. It’s fun to play with. My favourite workflow is teaching movements by first pupeteering motions then playing those…

Tab Clearing

Blog

Writing down quick takeaways/impressions as I clear up some of my reading backlog

LLM Steganography: Hiding Messages in Text

Video

Using LLMs to hide secret messages in generated text.

More=Better?

Blog

In this post I want to share a few quick experiments to show something that is both obvious and also perhaps underappreciated: you can get a sizeable quality boost from a…

Basement Hydroponics: Part 1

mini-hw-projects

I’ve been dabbling with hydroponics in my basement since discovering some unused grow lights from someone else’s abandoned attempt. It’s fun to see things grow! The system…

TIL: Intercepting Github Copilot with MITMProxy

TILs

We’ve been messing with integrating AI into notebook coding, and got curious about how Github Copilot structures the context it uses to generate code when called from a…

DIY Bio: A Cool Hobby, Not Quite For Me

mini-hw-projects

I’ve been dabbling with DIY biology, trying to modify some genes in ecoli and generally just messing about. Doing something like “making bacteria/yeast that glow under UV”…

Mini Experiment: Broken Multimodal Telephone

mini-projects

At a birthday party recently we played a game where a pad of paper is passed around a circle, with each person writing a sentence to describe a picture. The next person then…

Can LLMs estimate the length of a sequence?

mini-projects

While discussing another paper based around ‘LLM-as-a-judge’ style evaluation, the topic of length bias (longer responses tend to be rated higher) came up. This prompted…

Exploring 3D-Printed Instruments

mini-hw-projects

I’ve built up a bit of a collection of 3D-printed instruments:

Door Desk

mini-hw-projects

I’m slowly getting the basement organised into a nice workspace. In addition to my primary desk I wanted a standing workbench for soldering and other messy work. A scrap…

A few tips for working on high-surface-area problems

What do you do when there are too many pieces to fit in your head?

Blog

Some of my posts will now be duplicated to Answer.ai’s blog since I work there now. This one has just gone up (12 April) but since it has sat as a draft since January I’m…

TIL: GPT4-V Doesn’t Support Function Calling, but We Can Hack Around That

TILs

I want to use GPT4-V to evaluate images generated by a diffusion model and optionally modify the prompt based on the result. The problem is that GPT4-V doesn’t support…

Quick Experiment: Filtering Preference Data for Better DPO

blogs

With recent discussion around DPO (and proposed alternatives cDPO and IPO) I figured it was time we ran some experiments to see what works and what doesn’t. As a good…

Paperathon #1

Video

An experimental livestream reading through papers. A tidy version of the notes from the stream can be found here. Thanks to those who joined the live stream and helped out…

ZipLoRA: Any Subject in Any Style (deep dive and paper explanation)

Video

In this video we talk about merging LoRAs - the difficulties with a naive approach and the benefits of the new “ZipLoRA” technique. Paper: https://arxiv.org/abs/2311.13600

Mini Work Timer

mini-hw-projects

I lose track of time easily, so I’m trialling a little timer widget on my desk. Good for regulating Twitter breaks.

MLOps Community Mini Summit #3 Talk: What The Kaggle LLM Science Exam Competition Can Teach Us About LLMs

Video

Come for the LLM insights, and stay for the Llama-filled slides :)

Getting My YouTube Videos as MarkDown pages for this Quarto blog

TILs

Step 1 was getting the video info:

Evaluating Diffusion Models with PickScore

Video

Setting the scene for some future videos where I’ll explore ways to improve diffusion models through various tricks. Here we learn why evaluating diffusion models is hard…

How I ‘monetized’ an AI demo

Video

Creating a minimal template for charging users to run a custom AI pipeline.

Text-to-3D-Print

mini-hw-projects

Describe -> choose picture -> splats -> mesh -> sculpt -> slice -> print -> paint 😸

Mini Experiment: DoLa w/ a guidance scale

mini-projects

DoLa contrasts early and late layers of a transformer to elicit more ‘factual’ responses by looking at the contrast between the two. The theory is that earlier layers focus…

Mini Experiment: Hiding Easter-Eggs in LLMs

mini-projects

I just tried fine-tuning an LLM on a single sample to add an easter egg (Llama 2 7B + Platypus, using LoRA). I bet there are models floating around out there that have…

Gaussian Splatting explorations

Video

Let’s dive into Gaussian Splatting - what is it, how are scenes represented, and what fun things can we do with itThis is a fairly informal and code-heavy video - let me…

What is Speculative Sampling

Video

A quick explainer video for a technique called ‘speculative sampling’ or ‘assisted generation’ which speeds up language model sampling through the use of a smaller ‘draft’…

LLM basics #4 with the LLM Science Exam Kaggle Competition - Retrieval

Video

In this (delayed) final video of the series, we take a lightning look at one more useful technique to add to your submission arsenal - document retrieval.Finding the closest…

Exporting from Wordpress into a Quarto Blog

TILs

I have been blogging for years on a free Wordpress account, but figured I should finally migrate to a custom setup. I’m using Quarto, and in this TIL I document how to…

LLM basics #3 with the LLM Science Exam Kaggle Competition - Training a task-specific model for MCQs

Video

How do we adapt a model to solve multiple-choice questions In this video we dive into how AutoModelForMultipleChoice actually works, exploring the data processing and then…

LLM basics #2 with the LLM Science Exam Kaggle Competition - Generating Synthetic Data

Video

Another informal video as we work through the LLM Science Exam competition. Showing how you can - Start with a small dataset of high-quality examples- Train an LLM to…

LLM basics #1 with the LLM Science Exam Kaggle Competition - Zero-Shot approaches

Video

Talking about ways to use an off-the-shelf language model to solve a multiple-choice task. Covering – Intro to the Kaggle competition- Benchmarking with GPT3.5- Using the…

Exploring Softmax1, or “Community Research For The Win!”

blogs

Last week a guy called Evan Miller tweeted out a blog post claiming to have discovered a flaw in the attention mechanism used by transformers today:

Bark To Bark: Adding Noise Control to a Game

mini-projects

I found a game called ‘Russian Subway Dogs’ which is a ton of fun. I made a script to trigger the dog to bark in the game. Original Tweet with code link

Why and how I’m shifting focus to LLMs

blogs

While I’ve previously consulted on NLP projects, in the past few years my research focus has been chiefly on images. If you had asked me a few months ago about looking at…

A Recipe for Training Good Generative Models

blogs

Generative models are all the rage at the moment, and quality seems to be skyrocketing across the board. In this post, I share what I’m realizing is *the* key recipe that is…

Reward Hacking with RainbowDiffusion

mini-projects

An experiment incorpoarting an aesthetic model as part of the training loss went wring in all the right ways! And my boss let me release the result on huggingface :)

Style Loss Showdown

mini-projects

Reports

Remember Style Transfer? Turns out we can make it a lot better by moving away from Gram matrices! In this report, I dig into some alternative style losses including a…

Stylizing Video with Diffusion Models

Video

Reviewing existing techniques, introducing coherence guidance and running through the GEN1 Paper from RunwayML.

A Deep Dive Into OpenCLIP from OpenAI

Reports

This article explores an open re-implementation of OpenAI’s CLIP model for zero-shot classification and covers its additional uses and its potential issues.

InstructPix2Pix Explained - Edit Images with Words!

Video

Join me as I explain the core idea behind ‘InstructPix2Pix’ which let’s you edit images using natural language instructions!The thumbnail was generated with ‘Add…

Stable Diffusion Deep Dive Notebook Run-through

Video

In this video/notebook Johno shows us what is happening behind the scenes when we create an image with Stable Diffusion, looking at the different components and processes…

Genuary 2023

mini-projects

I did a few code doodles for Genuary 2023. Preview videos here.

Podcast E6 - Wasim Lorgat

Video

Wasim talks us through his approach to learning and building projects, touching on NBDev, building on big AI models like CLIP, his project ‘Meepo’, plans for building a new…

Mid-U Guidance: Fast Classifier Guidance for Latent Diffusion Models

Reports

Introducing a new method for diffusion model guidance with various advantages over existing methods, demonstrated by adding aesthetic guidance to Stable Diffusion.

DistilHN: Summarizing News Articles with Transformers

blogs

https://youtu.be/TBqzMfWpQrs

Building DistilHN - Using ML to Summarize News Articles

Video

Website - https://www.distilhn.com/Blog post - https://datasciencecastnet.home.blog/2022/12/30/distilhn-summarizing-news-articles-with-transformers/Code -…

HuggingFace Class

Video

Unit 2 - Fintuning and Guidance (casual notebook walkthough),Looking through the notebooks for the huggingface diffusion models class, unit 2 -…

HuggingFace Diffusion Model Class

Video

Unit 1 runthrough for the HuggingFace Diffusion Model Class

Editing Images with Diffusion Models (lit review _ overview of different approaches)

Video

How can we use diffusion models to edit existing images rather than generating completely new images1) Add some noise, denoise with a new prompt- VQGAN + CLIP paper -…

TGL Discussion Series - Hamel Husain

Video

In this video, I chat with Hamel Husain, a Machine Learning Engineer/Data Scientist passionate about making tools. Hamel is one of the most helpful and interesting people…

TGL Discussion Series - Jason Antic

Video

Jason is the creator of the incredible ‘DeOldify’ application for restoring historical photos, and one of the best experimentalists I’ve met. In this conversation we chat a…

TGL Discussions Series - Apolinario Passos

Video

Chatting with ML Art Engineer Apolinario Passos about keeping up with this fast-moving field, tips for getting involved, his journey to working on ML art full time and…

TGL Discussion Series - Teodora Szasz

Video

Talking with Teodora about her work as a Computational Scientist working on everything from medical AI to satellite-based archaeology to her most recent work on…

TGL Discussions Series - @EnzymeZoo

Video

@EnzymeZoo is an artist and developer. In this discussion, we chat about the community development work that goes into creating a tool like the Deforum notebook…

How Predictable: Evaluating Song Lyrics with Language Models

blogs

I was briefly nerd-sniped this morning by the following tweet:

Progressive Distillation for Fast Sampling of Diffusion Models (paper sumary)

Video

Having a bash explaining https://arxiv.org/abs/2202.00512 (and the follow-on paper https://arxiv.org/abs/2210.03142) I’ve been busy working on…

Summarising Neuromatch Deep Learning Course in 20 minutes

Video

An overview of the content covered by the NMA Deep Learning course. This material is freely available online, so if this video sparks an interest do go and check it out -…

Mapping Econimic Wellbeing from Space

Reports

Can you tell how wealthy a place is from space? It turns out that this is a very important question - an answer of ‘yes’ would mean that we can figure out how wealth is distributed within a country without needing to run expensive national surveys. For this project, I’ll be attempting to do exactly this, recreating some recent research to map poverty across Africa using existing household surveys and remote sensing data.

Paper exploration - Making Diffusion Models go BRRR!

Video

(Audio cuts for a minute but resumes at 4/55 - sorry about that)In this video we take a look at the paper ‘Elucidating the Design Space of Diffusion-Based Generative Models’…

AIAIART #9 - Evolutionary Algorithms and CPPNs

Video

CPPNs (not CPNNs as I say throughout the video!!) Compositional Pattern-Producing Networks! How embarrassing -)In this, the final lesson of AIAIART Part 2, we take a very…

WhistleGen V2: Making Music With Transformers

Reports

This article will dive into how the second version of WhistleGen works and explores how it enables you to generate folk music in ABC notation using minGPT.

Update Time

blogs

A few recent projects I’ve worked on have been documented elsewhere but haven’t made it to this blog. The point of this post is to summarize these so that they aren’t lost…

Fun With Neural Cellular Automata

Reports

In this article, we take a look at how to make pretty pictures (and videos!) using differentiable self-organizing systems.

AIAIART #8 - Neural Cellular Automata

Video

In which we discover the amazing world of Neural Cellular Automata, and see how to train them to match a style, a text description or an input image. Notebook link -…

CLOOB Conditioned Latent Denoising Diffusion Generative Adversarial Networks (GANs)

Reports

In this article, we’ll look at Image Generation with CLOOB Conditioned Latent Denoising Diffusion Generative Adversarial Networks (GANs), or CCLDDG, for short.

AIAIART Lesson 7.5

Video

Informal chat where we read through a few papers and look at a recent project. No central notebook for this lesson, but some resources we’ll be talking about – A brief…

AIAIART Lesson #7 - Diffusion Models

Video

Time to dive into diffusion models and see what is going on underneath the magic that is making the news at the moment -)If you’d prefer a slightly longer, more…

AIAIART Lesson 6

Video

AIAIART Lesson 6, diving into transformer models and their application to image synthesis. We’ll start by playing with text generation and build all the way up to creating…

Fine-tuning a CLOOB-Conditioned Latent Diffusion Model on WikiArt

blogs

Prompt: ‘A sunset landscape painting, oil on canvas’ (fine-tuned Wikiart model)

AIAIART Lesson #5

Video

Welcome back to AIAIART! This lesson recaps some of the core ideas from part 1 (lessons 1-4) and sets us up for the next few weeks, where we’ll look into some advanced new…

HugGAN sprint - demo project

Video

Come along for an informal look at making something for the #huggan sprint. We – Create a dataset and upload it to the huggingface hub- Train a model on that dataset- Work…

Sketchy Unet

blogs

The model demo running on Huggingface Spaces

Turtle Recall: A Contrastive Learning Approach

blogs

NB: A scoring glitch caused this approach to look very good on the leaderboard, but local validation and a fix from Zindi later confirmed that it isn’t as magical as it…

Come along for the ride as I dive into an ML contest (MABe 2 on AICROWD)

Video

A long rambly video as I pick a contest and put together a series of entries getting up to the top of the leaderboard. The first ~20 minutes is mostly checking out some…

AI Art Showreel (Jan 2022)

Video

A little taste of some of the things I’ve worked on over the past year -)More detail on many projects shown here can be found on my art twitter account…

AIAIART Course Retrospective

blogs

A small selection of images generated during the course

Playing with Tweet Sentiment Analysis

blogs

The average sentiment of the most recent 200 tweets from each country’s capital city.

AIAIART Lesson #4 (Live session recording)

Video

AIAIART lesson #4, the final lesson in this series. We go over some applications of the techniques covered in previous lessons and give ideas for things to try next.…

AIAIART Lesson #3

Video

Video run-through of Lesson 3 of the AIAIART course.

AIAIART Lesson 2

Video

Video run-through of Lesson 2 of the AIAIART course.

AIAIART Lesson 1

Video

Video for lesson 1 of the AIAIART course

‘Mars Affair’

Video

Music video created by me for David Ker aka @dadsalad2020. Made with GANs and love.

Getting to Grips with AI ART

Video

Taking a peek at some of the AI art stuff happening at the moment. How does it work, where can we go with it and what have I been working on. Sections -00/00 -…

Competition Attempts - ADDi Alzheimers Detection Challenge

Video

A quick overview of the ADDi Alzheimers Detection Challenge. Overview of the data, tips for improving your score and comparing models, visualizations of some of the cocks…

WhistleGen: Generating Traditional Irish music with ML

blogs

https://www.youtube.com/watch?v=B7SEP2p4Y1A

In Brief: Playing with Class Imbalance

blogs

We often encounter imbalanced data in the world of machine learning, and have to decide how best to handle this. In ‘real life’ it is up to us to decide how to evaluate…

BirdClef Entry: Bird Call Classification with FastAI

blogs

The Cornell Lab of Ornithology run an annual competition to identify bird calls in soundscapes. I decided to have a go at this year’s competition to get back into audio…

Competition Attempts - MABE Challenge first thoughts

Video

Documenting my first experiments as I attempt the MABE Challenge on AICrowd…

Competition attempts - BirdClef 2021 Initial Approach

Video

An introduction to the BirdClef Challenge and my first baseline entry. We – Look at the audio data- Introduce spectrograms- Look at ways to pick clips with high…

Meet DotSwarm - A Slime-Mould-Inspired Shader

Video

A webgl slime mould simulation.

Days of Code - Intro and Recap of the first few weeks

Video

Sharing the first few weeks of my ‘days of code’ project

Days of Code 2 - WhistleGen - generating Irish music with machine learning

Video

Generating and playing some ‘traditional Irish style’ music with code. This video covers scraping the tunes, training a model and turning the output into music. Try it…

Language Models for Protein Sequence Classification

blogs

A 3D model of a protein kinase

Data Glimpse: Predicted Historical Air Quality for African Cities

blogs

Air quality has been in the news a lot recently. Smoke from fires has had thousands of Californians searching for info around the health hazards of particulate matter…

Personal Metrics

blogs

This is just a quick post following on from some recent conversations in this area. tldr: Tracking some data about yourself is a great exercise, and I highly recommend it.…

Update: Why the pause?

blogs

It’s been nearly four months since my last post on this blog. I’d like to get back into writing things up here, but before putting up another normal post I figured it was…

Self-Supervised Learning with Image网

blogs

Until fairly recently, deep learning models needed a LOT of data to get decent performance. Then came an innovation called transfer learning, which we’ve covered in some…

Meta ‘Data Glimpse’ - Google Dataset Search

blogs

Christmas came in January this year, with Google’s release of ‘Dataset Search’. They’ve indexed millions of cool datasets and made it easy to search through them. This post…

Swoggle Part 2 - Building a Policy Network with PyTorch, dealing with Cheaty Agents and ‘Beating’ the Game

blogs

In part 1, we laid the groundwork for our Reinforcement Learning experiments by creating a simple game (Swoggle) that we’d be trying to teach out AI to play. We also created…

Swoggle Part 1- RL Environments and Literate Programming with NBDev

blogs

I’m going to be exploring the world of Reinforcement Learning. But there will be no actual RL in this post - that’s for part two. This post will do two things: describe the…

Behind the scenes of a Zindi Contest

blogs

Snapshot Serengeti - Working with Large Image Datasets

blogs

Driven Data launched a competition around the Snapshot Serengeti database - something I’ve been intending to investigate for a while. Although the competition is called…

Deep Learning + Remote Sensing - Using NNs to turn imagery into meaningful features

blogs

Every now and again, the World Bank conducts something called a Living Standards Measurement Study (LSMS) survey in different countries, with the purpose being to learn…

Zindi UberCT Part 3: Uber Movement

blogs

Uber Movement has launched in Cape Town

Zindi UberCT Part 2: Stepping Up

blogs

In part 1, we looked at the SANRAL challenge on Zindi and got a simple first submission up on the leaderboard. In this tutorial I’ll show some extra features you can add on…

Zindi UberCT Part 1: Getting started

blogs

Welcome to the first in a three-part series on Zindi’s Uber Movement SANRAL Cape Town Challenge. This tutorial will take a look at the challenge, start exploring the data…

Packaging a classification model as a web app

blogs

My shiny new web app, available here

Pothole Detection (aka Johno tries fastai)

blogs

This week saw folks from all over the AI space converge in Cape Town for the AI Expo. The conference was inspiring, and I had a great time chatting to all sorts of…

Trying Automated ML

blogs

Some students had asked me for my opinion on automated tools for machine learning. The thought occurred that I hadn’t done much with them recently, and it was about time I…

Mapping Change in Cropland in Zimbabwe (Part 2)

blogs

In Part 1, I showed a quick way to get a model that predicts cropland extent, using someone else’s model as a starting point. This was a fun exercise, but in today’s post…

Mapping Change in Cropland in Zimbabwe (Part 1)

blogs

I’m going to be working on a project that will ultimately require manually outlining tobacco fields in Zimbabwe. To help locate potential fields, it would be nice to have a…

Data Glimpse: Cropland and Settlement maps from QED.AI

blogs

The point of this Data Glimpse post is to feature a wonderful yet badly publicized data source: https://maps.qed.ai/. Using crowd-sourced data, they built really accurate…

Data Glimpse: Nighttime Lights

blogs

This ‘Data Glimpse’ post will look at the Global Radiance-Calibrated Nighttime Lights dataset [1], available through Google Earth Engine. However, the method shown here can…

Data Glimpse: South Africa’s Hydrological Data

blogs

South Africa’s Department of Water Affairs (DWA) makes all kinds of data publicly available through their data portal: http://www.dwa.gov.za/hydrology/. The download…

Data Glimpse: Visualizing Economic Activity with the G-Econ Project data

blogs

This is the first ‘data glimpse’ - a short exploration of an existing dataset, with code and examples showing some of the ways the data can be used. For today’s glimpse…

Tutorial: Improving Crop Type Predictions

blogs

Following on from the last tutorial, this post will look at some ways we can improve our crop identification method. At the end of the last post, we were using a CART…

Tutorial: Predicting Crop Types with GEE

blogs

Zindi is currently hosting a competition to classify fields by crop type using Sentinel-2 satellite imagery. They provide labeled fields with the crop type, and a separate…

New Database: Forest Change in Different Regions

blogs

Forest loss is a major problem facing many parts of the world right now. Trees are being cleared to make way for agriculture, or simply cut down for fuel and timber.…

Zindi Competition 1 - Making Art!

blogs

I’m going to try entering some Zindi competitions this week. First up is the ‘AI Art’ contest. I have many crazy plans, but my nascent tensorflow skills mean everything…

Zindi Competition 2 - Trying CatBoost on the Traffic Jam Challenge

blogs

Zindi ran a challenge predicting bus ticket sales into Nairobi. It is now closed, but we can still make predictions and see how they would have done. This was a very quick…

Looking at traffic/congestion vs air quality AKA a quest for data

blogs

I’m currently on a mission to explore the different datasets available publicly, and hopefully to add to that list. One key reason I’m passionate about this is that data…

ML and IR Tomography

blogs

I studied Electrical and Computer Engineering at UCT, and the final year project was my chance to really dive deep into a topic. I chose IR tomography, and explored various…

Mapping Baobabs, Part 3 - Model Applicability

blogs

In the previous two posts, we built a Species Distribution Model and used it to predict the density of Baobab trees in Zimbabwe. Then we tried some more complex models…

Mapping Baobabs, Part 2 - Qualifying Model Performance and More Complex Models

blogs

Update, July 2020: GEE now has a good random forest implementation (ee.Classifier.smileRandomForest) that can do regression - I’d suggest using that instead of the approach…

Mapping Baobabs, Part 1 - Modelling the Density of Baobab Trees in Zimbabwe with a Linear Model in GEE

blogs

This is the first in a multi-part series exploring Species Distribution Modelling (SDM) with the Google Earth Engine. In this post, we’ll take a look at the data we’ll be…

Christmas games - simulation and lazy stats

blogs

This Christmas, I was introduced to several new games by my new extended family. Much fun was had making up new games and rediscovering old, but one game annoyed me…

Curious correlations

blogs

I wanted to write this up to show how easy it is becoming to test ideas and find interesting trends in data. Please don’t draw too many conclusions from the numbers here -…

init(self): What is this blog

blogs

CIRTS - Configurable Infra-Red Tomography systems

mini-hw-projects

Using cheap IR LEDs and phototransistors to understand tomography. Using machine learning to make inferences based on sensor data. Etc :)

Owning a 3D printer

mini-hw-projects

I got my first 3D printer in 2018. It was a Creality Ender 3. I printed a lot of parts for my own projects but also paid off the cost of the printer by printing parts for…

Studying automation with an octopus overlord!

mini-hw-projects

It’s easy to automate machines, harder to automate myself. So I made this - a glorified countdown timer keeping me focused on studying :)

Mini PCB printer

mini-hw-projects

Melting toner (as etch resist) directly onto copper clad board using a laser moved by a DVD drive XY system.

Scanning EIT Imaging System (SEITIS) Attempt

mini-hw-projects

A failed attempt to make a scanning electrical impedance tomography system. Built from old printer parts, I had the motion stage (jankily) working and some of the…

Pimp my dormroom

mini-hw-projects

Ah yes, what every dorm room needs is cardboard furniture, DIY storage and tons of tools. Good memories :) More pics at https://hackaday.io/project/9372/gallery.

Cell Phone Oscilloscope

mini-hw-projects

A preamp to use a cellphone mic input as a basic oscilloscope. Protects the phone from nasty voltages. <40V, <20kHz.

Micromouse attempt

mini-hw-projects

I tried to build a micromouse, but let the magic smoke out the day of the competition. As was my style at the time the guts were a bare bones ATMega328 that I’d got as…

Kwese - musical instrument

mini-hw-projects

Capacitive touch-based instrument / midi controller. Similar to a makey-makey. Created for a hackathon. My first introduction to the brutal reality that you can’t in fact…

Electronic Bagpipes

mini-hw-projects

I made an OpenChanter-based electronic bagpipe, with capacitive touch on screws where the holes would typically be.

Mini Laser Cutter

mini-hw-projects

Tiny laser cutter built out of broken DVD drive parts.

Multitouch Surface

mini-hw-projects

Whipped up a quick DIY multitouch thingee - really cool to play with and cost $0 (not bad considering Microsofts commercial alternative costs $12000!). Uses an open source program called CCV to process the video feed from my IR camera in the bottom of a box. The box is covered by a piece of perspex and a sheet of paper. By analysing the shadows on the paper the fingers can be detected. I also set a projector in the box so that you see the screen react as you touch it.

DIY Cellphone Macro Lens

mini-hw-projects

Found an easy and free way to take amazing macro shots with your cellphone using the lense from inside a DVD player. Pics and long explanation in ‘more details’ (warning, large ~2MB images)

Laptop in a briefcase

mini-hw-projects

After I managed to get my hands on an old laptop, I wanted to try my hand at making it into a “picture-frame” display, as seen all over the web. However, with the power of a (n admittadly derelict) laptop at your disposal it seems a bit of a waste if all you want is a slide-show. In addition, my mother didn’t look too happy with the idea of me liberating a frame from one of her pictures.

Junk VDG

mini-hw-projects

I threw together a Van de Graaff generator out of stuff lying around. It worked, but the whole point was that by doing everything wrong I can now explain how it should have been done. Check out the full article for the details.

Easy 3D Scanner

mini-hw-projects

How cool would it be to be able to scan objects in 3D? Turns out its not hard, and can be done with stuff you probably have lying around the house. I made mine using a cheap laser pointer, a broken wine glass stem (sorry mum :) and some cool software -makerscanner. This is a piece of software designed specifically for their design, but I will explain how to adapt it if you don’t have a 3D printer. I also usedMeshlabandBlenderto clean up the results.

IR Webcam

mini-hw-projects

When I found out this was possible I couldn’t resist having a go. Sadly, I didn’t document the build process but there are many tutorials out there already explaining all the details. A short explanation and some more pics in ‘more details’.

Title

mini-projects