visekr

by the seashore?

No biometrics collected. It's only scanning bodys, clothes and details on gear.

yes! although the techniques aren't perfect.

I'm using a YOLO-WORLD-XL object detection model. Lets me detect objects using text. This is the initial filter that scans for agents - once those are detected and outlined with bounding boxes the entire image and each cropped bounding box are then sent to chatgpt to confirm if the image looks legit. Once image passes those checks - I create image embeddings of each agent using CLIP and those are stored in a vector DB, and each agent is then compared to the DB and matched.

The matching system isn't perfect - but I think good enough to get the point across and can be easily tuned with more data! Happy to take suggestions here - I just spun this up over the weekend

i was gonna say the same - as my post got sunk into oblivion after hitting second page

Loveee that idea.

Would be cool to even have designated groups so you could filter by left wing, right wing, international, etc.

Gonna work on this after I optimize the codebase

of course! WorldNewsAPI - wasn't my first pick but was the most affordable for me, but i like it so far, no complaints!

ah thanks : )

yes - not sure what detail you want but I'll try to be thorough here.

Tech stack is React, Next JS, Radix UI lib, Tailwind, Firebase for database auth and cloud functions, Vercel for deployment, OpenAI for article classification, WorldNewsAPI for news.

Cloud Functions run at different intervals, I have one that runs every 30 minutes (recent news), 12 hours (weekly news), 84 hours (monthly news) and 168 hours (quarterly news). This is a big WIP

On a cloud function run it fetches articles from newsAPI, this search is done with a mix of keywords that determine if its related to the current administration, then gets filtered down to 'credible' sources (listed in a comment on this thread). The content of the articles that pass these filters get send to gpt-4o to be classified for relevance (is it an action of the administration / related) if it passes that >70%, the ai determines if it matches any of the 14 steps of facism, and will add a direct quote / impact / summary of why it matched that step, and finally will get added to DB and displayed on the site.

In terms of development - heavy use of Cursor IDE, and start with v0 by Vercel which is why I was able to get this up and running within ~3 days. Although the codebase looks disgusting rn

I'd love to tap into that eventually! Right now I'm using an API that give me a dump of like all articles and I filter that through a list of 'credible' sources. I'm not the most educated about what are good sources for this but this is my current one. (it also shows on the site under - 'what is this') Would loveee feedback on this list.

Allowed Sources Major Wire Services (1.0) Reuters, AP, AFP, Bloomberg Top US News (0.9) NYT, WaPo, WSJ, ProPublica

Conservative Media (0.7-0.8) Fox News, National Review, Washington Times, The Dispatch, Washington Examiner, RealClearPolitics, Reason

Progressive Media (0.7-0.8) Mother Jones, The Nation, Democracy Now!, New Republic, Jacobin, Truthout, Slate

International (0.8-0.9) BBC, Guardian, FT, Economist, DW, France24, Al Jazeera

Fact Checkers (0.9) FactCheck.org, PolitiFact

Major Networks (0.8) NPR, NBC, CBS, ABC, CNBC

Digital Media (0.7-0.8) Atlantic, Wired, Axios, USA Today, Time