Already Created
Last updated
Last updated
While driving down the road I got stuck in traffic. I thought this happens every single day, how come the map doesn't account for it. I thought what happens if I were to map the accidents and comb through the data to find a pattern.
Grab the data of where accidents have happened in the past and incorporate that into a pathfinding to determine how to route people away from it before it happens.
While playing rocket league one night a debate broke out between which is better, having a Dog or a Cat. I was in the process of creating TOBIAS and thought it would be cool to say "Hello cat". Later on down the line, it would actually remember cat/dog based on shape.
To detect if there is a dog or cat in the picture. The real twist that I wanted to implement was figuring out the breed of the animal but that was very inaccurate due to the amount of information was given in an image.
I just got out of an extra credit lecture talking about some advanced things and I wanted to do a project. I thought, what happens if I could create something that could predict what sounds would be good. That seemed impossible so stepping back I thought it would be easy to start out by just classifying the songs.
It would take in a song or part of a song and analyze it from already known songs in genres and classify it. Based on nothing being perfect I would set the percentage of similarity around 85% and if it is not that in 1 category it would be unknown.
So my friend says he likes cars but I feel like that is a lie because he doesn't know some of the most famous cars out there. Granted they could be old but you should know them. My idea for this program had 2 things in mind. 1. It could find cars on the road while driving or a picture 2. Classify the make and model of the car for later search. The reason why is no one can remember all the cars (unless I guess someone that is autistic or a genius).
Detect where a car is on the screen. Count the number of cars. Classify each car to make and model. Simple and easy. Also, this could be useful for traffic as more cars mean more traffic (for the most part).
The reason is there is already technology out there to do this. Also, the name of classify and count is more for classifying the number of axels on the car. Not going into much detail because more can be found on the products sold and looking into what captures all this data.
It was already created through an API and don't think I could match this speed at all
I have sooooo many people wanting to get into stocks thinking, it will make me rich, or something like that. I was taking a Machine learning class at the time and thought, what happens if I could do that. Of course, it would take lots of time but what the heck it is possible. I mean even if it is not perfect, 85% of the time is better than trying to do it on your own. It seems like get rich quick but what is wrong with that if you know how to code well.
Basically, the program would take in past data and trends and try and predict which investment would be the best. This is probability-based so we aren't looking for the biggest difference in price, but instead the percentage of change and confidence in that change through Machine Learning.
So a little thing I used to do in school was using the tools the teachers used against them. What I mean by that is self-explanatory with the title but if you don't get it I guess I will break it down more. Say a person submits a document to a teacher and most commonly the teacher would run it through a plagiarism checker. Well, I would run the question sheet the teacher gave me and it would search more efficiently through the internet for the answers instead of typing each question individually into google. Automating the boring things.
Using some API and NLP to check the plagiarism of text. I decided the online programs were good enough as a person would only run the program once per document and if more documents it is still more efficient than running each question through google.
I have had multiple times when I need to grab information from the internet or crafting a specific request for a website. That is why it is good to have that code.
This takes my original thought of sending a request and processing what is returned in every programming language.
I am making the project because I am bored and waiting on my job to start. My inspiration for the project was taken for when I was pen testing and thought how many invalid URL with error code 400 show up. Because of that I decided why not send them a message about it
The program just does crawl, which is good reference, but other projects will include much more than that.
My girlfriend and I were trying to come up with a movie to watch and she asked for the ratings when pulling them up it was very poor. I then went searching for my other favorite movies which were rated poorly.
I want to create a system that can predict how successful a movie will be when it is released. Not only that but then take that information and see if you can predict how much the movie will make.
In meeting and heard and was wondering if could make it like a madlib speeding up the process of creating a cover letter.
Someone already did it. Not the best but it gets the job done
After looking for a job sometimes, I noticed that a common trend was happening for some successful people tailored the resume to each company. I didn't have time for that so of course, I was going to go to automation because it most likely gave me the best odds which I would like.
Building a resume using user input to help with the process
I thought it was a good idea until I thought into it more and websites that create resume are good enough and look really good for format. To even compete with that I would need a lot of formats and a lot of extra stuff that in the end would not be worth it.
I struggle with interviews to be talking to a headhunter. I was having a tough time understanding their accent which was a constant struggle, so I was wanting to just remove it. How hard could it possibly, be.
It would take out the accent potentially to text and then a voice. The real thing to make difference would be the original voice, if possible, just refined.
As I was drawing up an outline for this idea, I realized overall that it already exists. When you break it down, accent is basically what differentiates a voice from another voice. Removing that and you get machine sounding and the Process would be audio -> Text -> audio. Which is just repeating the speech to text.
I was going through and on linked in I saw someone doing an interactive game that would move the balls based on where his arms were. That seemed useless to me, and I remember the Microsoft HoloLens commercials so it would be much more useful to draw something in the air and save it for later editing.
The program would use the built-in computer camera or an external one, really whatever it can find. It will find the hand and wherever it is pointing it would draw like on a canvas. These are strictly 2d drawings that would be saved in JPEG mode.
I was working on the TOBIAS project and realized that I was hard coding every single command to be understood. I didn't want to do this every single time for the rest of the project if I were to grow. Another benefit of a chat bot configuration would be the more realistic feel of the software.
No need as the TOBIAS Project was personal thus coming to an end.
TOBIAS has the weather because every personal assistant should be able to tell the weather at the current location. It is through the API and requires an API key, which I find really annoying. I want it to be based on web scraping and not reliant on getting an API key.
Grab the weather information (Without an API key)
Not a valid project anymore because TOBIAS is being discontinued.
I was talking about interviews and people said that posture was a huge thing. I took that into account and was wondering if posture said more than just what someone was feeling.
Detecting the Human Pose and create a feature around those components
My handwriting is not the best and took notes but want them searchable so I kind of want to be able to scan notes and search the info during a test. Totally would let me have bragging rights.
Turning handwriting to text through Neural Network
It is literally one of the baseline projects that are done in Deep learning. Can find it on any starting up website. Stupid to think this would be hard.
