yea yolov7's .pt is fine but we're not getting the best Performance™ out of our hardware. so in this tutorial We will be converting yolov7's .pt to Nvidia's TensorRT .trt format. there won't be much complex janky setup , as it is a streamlined with using a docker container (you do have to deal with nvidia drivers though but you're probably already dealing with it anyways)
For this demo we will learn how to find the distance from a given longitude/latitude coordinates to another longitude/latitude tagged place of our interest.
this project is under active development, things are subjected to changes! for the past few months , i've been doing a few project with the YOLO framework, in my free time. Now, there is a good platform for labeling, and generating datasets called roboflow, but due to the amount of data that i've accumulated, it supasses the amount allowed on the free tier of roboflow , which is 50000 generations. So i've decided to write my own conversion and augmentation tool called kesa. Currently kesa has four commands.
it's been awhile since my last post (it wasn't even complete lol) , but during that time i've been tinkering around with various deep learning models and frameworks, and what i've found out is that augmented data is important in order to train accurate ML models, while there are libraries made for augmentations like albumentations readily available , the type of augmentation that i want to do is not avialable anywhere (if there is a library that can do this ,please do tell). what im trying to do is taking a labeled image from labelme and then putting it in a different background, so let's get to it!
edit : im too lazy to finish this article smh disclaimer: there is alot of mentions of penis in this blog post
