GeoTrainr Used AI

deploying and adding a progress bar

i literally spent all this time trying to deploy this and finally got it!

ive literally been trying for HOURS to deploy this thing but it takes up too much memory for any free services like im actually crashing out

working on deploying and small adjustments

working on deploying and frontend edits

working on reducing time to create maps and adding download map feature

setting up a faq section. tried to do a component first but then learned that existed lol

okay i create a function that creates a json file that geoguessr can read, and I tested it and looks like it works! (somewhat). The location spread is really weird and not all locations are within the selected country. also to go through 500 locations takes 10 min which is a bit too long, so I need to figure out how to make that shorter. but all things considered, pretty good progress so far

HOLD UP SOMETHING MAY HAVE WENT RIGHT

currently working on applying my ml model to filter out locations to include in maps. also trying to figure out how to do this the most effectively without downloading images, taking up too much space, etc. sort of just doing trial and error and debugging along the way

i set up the backend with a flask app and now the frontend posts the user's data to the backend. next step is figuring out how to connect this data and the ml model

frontend!! I worked on this for about an hour and a half, so i lowkey forget what the other hour was for ngl. but for the frontend i did the design and interactive adding/deleting functions of user weaknesses. next step is to send that data to a backend prob flask app

dataset is now at 1000 images! retrained and ran some test detections

now have a system that sorts through data for me. ill now add to my dataset

trained and tested the model locally using yolov5. now need to collect more images with only bollards using the model to add to my dataset to make it more reliable

i trained a model on roboflow, but now trying to retrain a model locally as I realized roboflow would cost a lot. im going to use yolov5, and I havent done any machine learning stuff like this before so lets see how long it takes lol

I picked Mapillary and its API to retrieve images to help train my model. Right now, I am just focusing on bollard object recognition as a test round and labelling some data.