Hi, I understand that many folks are trying to upskill in computer vision and hence are maximizing knowledge competitions such as this.
I've decided to open-source my current notebook of 0.961+ to get you started as a benchmark, I've also done some explanations of my tricks & tweaks so you can follow through.
please find the notebook attached here: https://github.com/osinkolu/Road-Segment-Identification-Zindi
hoping to see you all above me on the Leaderboard, but more importantly that you gain some insights✌.
Nice approach @Femi 👍