Neurasight NN - Machine Learning and Computer Vision

Finally got the page working, published it on github repositories and hugging faces for the backend, if you upload an image of a brain MRI it will scan it for one of the three classes of tumor, or detect if theres no tumor. Had to make a docker file to upload it to hugging faces. The model still needs some training for better results but it works thats the important part. If you have any feedback please let me know so I can work on it and I hope I make a second version of this project :)

Got a first version working which classified in no tumor and tumor in the page but forgot to take a picture, I went with another model of CNN which will have better acurracy and also will put a square around the tumor. Rn im training the second CNN model as you can see on the image below.

Hi so, I worked for a long time on the connection with JS and python, the moment that Im publishing this devlog is the moment the cnn is getting trained, it will take about 15-30 min to finish training, so i’ll talk about what I made.

So I activated a virtual enviroment (.venv) and installed a lot of dependencies, I thought about using an already trained cnn but I decided to train my own. I used tools such as python, .venv, flask, opencv, numpy, scikit-learn, joblib and well the kaggle dataset. Rn the training and processing of MRI images is being held by the model. Once it finishes training we will connect the model to a Flask API and integrate it with the web interface for real-time predictions. I would like that you give me feedback of how I can make it easier Also you can take a look at how the web interface turned out., in my opinion it look reallyyyyy good

Finished most of the html and css look, it looks a lot better now, i will now finish tha page with the js and get to the important part, creating and training the ai model.

Added an image related to computer visión so its easier to understand what the model is doing.

Added a blur effect on the background and made it more Dynamic, it now has a 0.8s transition that make the text and everything appear, some border-radius yk :p also has a hover effect on top of the cards. Also the analyze image button now has a different look and a hover lol.

Added some color to the page, added some information to know about the Project, added some buttons that will let you open your folder and put an MRI image, also separated it by sections and on top we have a navigation bar. Next I’ll add some CSS so it looks better.

Hi guys this my devlog, first of all I Will explain the Project and the first thing I did. I wil be using CNNs to detect patterns in MRI scans and classify brain tumors. First of all I downloaded the dataset Brain Tumor MRI Dataset from kaggle. It has 4 clases, glioma, meningioma, notumor and pituitary. I started watching some of the photos that come with this folders and next I Will start doing an HTML page with css and js so that you can use it too. Also I wWill work with Python, flask and other things you’ll see later in the devlog.