stephanst

4 ms·

1.
Show HN: VoxHearth / small local privacy-focused STT Mac app(github.com)
1 points by stephanst 1mo ago|0 comments
2.
Show HN: Tator the local ML image annotator(github.com)
2 points by stephanst 1y ago|0 comments
3.
by stephanst 2y ago
new dataset, new classes and model is yolov8 because most people prefer it. Also open weights.
4.
by stephanst 2y ago
Yep exactly. Commercial satellite inagery isn’t great for detecting anything the size of a tank (yet) but drones will get you there. The real difficulty for detecting anything like this is the training data, IMO that’s the interesting part
5.
by stephanst 2y ago
generalisation is 90% of the problem yep. This model is trained with a lot of augmentations and quite diverse data, but it still really needs fine tuning to specific use cases to get great efficiency unless you don’t care about false positi
6.
by stephanst 2y ago
you don’t need AI for that, classic computer vision will work fine ;)
7.
by stephanst 2y ago
it is being used a lot for disaster response and traffic monitoring yep. Also surprisingly to me for analytics on utilisation of all kinds of infrastructure y trading firms. One of the bigger motivations to make it free was to see where peo
8.
by stephanst 2y ago
Some of the design choices of YOLOv7 make more sense to me in the choices of default augmentations and the structures of the very large versions of the networks. I find I can push it to marginally better recall. It’s slower than Ultralytics
9.
by stephanst 2y ago
that class never really worked and has been removed from the new version of WALDO FYI, it’s not a military thing and shouldn’t be used as such
10.
by stephanst 2y ago
this was >50% of the motivation
11.
by stephanst 2y ago
Hey, thanks for posting. New release is coming tomorrow on HF BTW. AMA
12.
Show HN: I built an open-source AI system for drones(github.com)
5 points by stephanst 3y ago|1 comments
13.
Show HN: Talk to All of Wikipedia with ChatGPT(wikipediagpt.app)
2 points by stephanst 4y ago|0 comments
14.
by stephanst 4y ago
looks cool, thanks !
15.
by stephanst 4y ago
I’d love to try it ! TBH this app was made in less than a day as a proof of concept: my real work focus is on AI for drones and satellite imagery. If you’d like to build an alternative with Qdrant I’d be happy to merge it into the repo!
16.
Show HN: GPTflix: a simple framework for the OpenAI and Pinecone stack(github.com)
4 points by stephanst 4y ago|5 comments
17.
by stephanst 4y ago
Hey HN people, I'm releasing the source code of my little streamlit app called GPTflix that allows you to interact with >450K movie reviews and plots using vector search and the OpenAI LLMs. Enjoy!
18.
by stephanst 4y ago
url forwarding is broken, you can use https://www.gptflix.ai
19.
by stephanst 4y ago
And... it's back online, supported by Ben's Bites, my favorite AI newsletter! https://bensbites.beehiiv.com/
20.
Show HN: Blowing $100 on OpenAI in a few hours with context injection(gptflix.ai)
5 points by stephanst 4y ago|3 comments
21.
by stephanst 4y ago
Sorry it appears that HN doesn't accept the format of the link somehow... it should be here https://gptflix.streamlit.app/
22.
by stephanst 4y ago
Hey HN. I made this last week for fun. It's using Pinecone as a vector database / backend, and context injection + data retrieval via embeddings search to power the chat function! It will only answer about movies that it knows (ab
23.
GPTflix –> ChatGPT to help you choose what movie to watch(share.streamlit.io)
2 points by stephanst 4y ago|3 comments
24.
by stephanst 4y ago
Hey HN. I made this last week for fun. It's using Pinecone as a vector database / backend, and context injection + data retrieval via embeddings search to power the chat function! It will only answer about movies that it knows (ab
25.
Show HN: Waldo alpha 0.1 / free embedded AI for UAV-based detection(github.com)
1 points by stephanst 4y ago|0 comments
26.
Show HN: Waldo – free AI for UAV safety
1 points by stephanst 4y ago|0 comments
27.
Curious ML dataset for detecting Russian suicide drones(universe.roboflow.com)
2 points by stephanst 4y ago|1 comments
28.
by stephanst 4y ago
The dataset is a bit of a mess: there are labels as well as masks and they are not split up correctly... but still looks pretty usable with a good augmentation stack.
29.
by stephanst 4y ago
42 image batch size is scaled to the input layer of my DNN, otherwise yes :)
30.
by stephanst 4y ago
the 90% / 10% image was not a real world stat, it's supposed to be an illustration... but not well worded I'll grant you that. What you are doing is the same type of work that I would put in the "pro user" bucket li

More ›