Dayo

SaaS Developer | CRM • Funnels • Automation
Building a nanny-matching platform for Nigeria
Open to Freelance • Contract • Full-time
DM open

[Dayo @Dayhur]

Can you spot what the AI missed? Red boxes = human labels. Blue boxes = my model’s predictions. The baseline finds obvious road damage but misses some smaller/distant potholes. Next: test it on Lagos roads. #BuildInPublic #ComputerVision

[Dayo @Dayhur]

your future clients and employers are watching your page. Post your work, not just your thoughts.

[Dayo @Dayhur]

Reminder: slow progress is still progress. One hour of coding today beats zero hours waiting for motivati

[Dayo @Dayhur]

@aksenHQ I’ll add real Lagos images of small/distant potholes with consistent boxes, varied lighting/weather and look-alike negatives. I’ll test on held-out routes. If a pothole is only a few pixels, I’ll treat it as beyond reliable detection rather than label noise.

[Dayo @Dayhur]

First baseline done for my Lagos pothole project using Aegis’s 1,482-image public dataset. YOLO26n, 5 epochs. Test mAP50: 0.851. It detected obvious damage but missed some smaller/distant potholes. Next: Lagos road tests. #BuildInPublic

[Dayo @Dayhur]

name your files and folders properly today. Future you will not remember what "final_final2" means.

[Dayo @Dayhur]

Nobody is coming to give you an opportunity. Build it small, show it publicly, and people will start to notice.

[Dayo @Dayhur]

Tip: if a problem takes you more than 30 minutes, take a short walk. Your brain keeps working and the answer often shows up.

[Dayo @Dayhur]

@progerallworld That’s a great point—thank you. For my Lagos dataset, I’ll split by street/route or video, not by individual frames, so similar frames from the same road segment stay on one side. That should give a more honest test of generalization.

[Dayo @Dayhur]

I’m starting a Lagos pothole-detection project. I’ll test whether AI can mark possible potholes in road images, then record their locations. I’ll share results, mistakes and lessons as I learn. #BuildInPublic #ComputerVision

[Dayo @Dayhur]

@aksenHQ I’m starting with the Pothole Detection dataset by Aegis on Roboflow Universe—1,482 images, CC BY 4.0—for my baseline. Later, I’ll test on Lagos road images and build a local dataset for fine-tuning.

[Dayo @Dayhur]

Small wins count. Fixed a bug, finished a lesson, sent one message to a client? That’s progress. Be proud of it.

[Dayo @Dayhur]

Be honest: what’s the one thing stopping you from posting your work online? Fear, time, or “it’s not good enough yet”?

[Dayo @Dayhur]

You don’t need 10 skills. Pick one, get good at it, and people will pay for it.

[Dayo @Dayhur]

Do the hardest task first today. Everything else feels easy after that.

[Dayo @Dayhur]

Monday. Light is on, data is loaded, laptop is charged. Let’s get to work

[Dayo @Dayhur]

you don't need a perfect plan for the week. You just need to show up tomorrow and do the first task.

[Dayo @Dayhur]

Quick tip: pick one thing to learn this month and stay with it. Jumping between five courses is why most people feel stuck

[Dayo @Dayhur]

New month, November 1st. Write down one goal you want to hit before December. Just one. Drop it below

[Dayo @Dayhur]

Writing down what you learn forces you to understand it. People also correct you and teach you in the comments.