Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: What problem are you close to solving and how can we help?

news.ycombinator.com

Post reply on HN
Print view View original

211–220 of 486 posts

Re: Ask HN: What problem are you close to solving and how can we help?

#211

Post by ad404b8a372f2b9 »

needz wrote: post #203
ad404b8a372f2b9 wrote: post #158

Earlier quoted context omitted.

Have you tried an off-the-shelf solution like Tesseract? It works quite well if you do the recommended preprocessing.

The preprocessing suggestions I see are to crop out everything except for the numbers and I don’t know how to do that programmatically. There’s many kinds of displays: rollers, 7-segment, dot matrix, and LCD. The preprocessing to increase DPI to 300 did not help when I tried Tesseract, unfortunately. It’s hard to achieve a good contrast between the numbers and the backdrop

There are a lot of other options and preprocessing methods you can use to get better results. It's hard to tell without seeing the picture but thresholding/binarization might help with the contrast. In order to isolate the text, the mode option also makes a lot of difference: https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html#...

If that doesn't work you'll have to add a text localization model to your pipeline.

Re: Ask HN: What problem are you close to solving and how can we help?

#212

Post by bob1029 »

red0point wrote: post #194

I‘m trying to re-/sell cheap bulk object storage, by renting cheap dedicated servers (e.g. Hetzner), connect them using 10GbE and putting them into a big Ceph cluster. My problem is how to bill people for consuming object storage properly. Do you do it retrospectively and take the fraud risk? Are there any pre-existing platforms that do Ceph billing?

Can you explain how the economics of what you are doing would be competitive with something like S3 or B2? I feel like there could be a market/margin here but there are a lot of numbers involved to figure out the specifics.

Re: Ask HN: What problem are you close to solving and how can we help?

#213

Post by ufmace »

k1rcher wrote: post #2

We are having atrocious READ/WRITE latency with our PG database (api layer is django rest framework). The table that is the issue consists of multiple JSON BLOB fields, with quite a bit of data— I am convinced these need to be abstracted to their own relational tables. Is this a sound solution? I believe it is the deserialization in these fields of large nested JSON BLOBS that is causing latency. Note: this database…

I don't think the latency issues are necessarily related to the poor schema. I'd say to dig into the query planning for your current queries and figure out what's actually slow, since it may not be what you expect.

Rearchitecting the schema might be worth doing. From the technical side, PG is pretty nice about doing transactional schema changes. I'd be more worried about the data though. Are you sure that every single row's Json columns have the keys and value types that you expect? Usually in this type of database, some records will be weird in unexpected ways. You'll need to find and account for them all before you can migrate over to a stricter schema. And do any of them have extra unexpected data?

Re: Ask HN: What problem are you close to solving and how can we help?

#214

Post by tmilard »

kroltan wrote: post #205
tmilard wrote: post #16

https://free-visit.net : Like Matterport but with an FPS game engine. I am looking for my first client: Ideally someone in charge of a Museum/gallery or other grandiose indoor space.

I reiterate the sibling comment about mouse control, either use the actual mouse locking APIs of browsers, or make drag feel like drag. Additionally, I think it would be best if by default you were stuck to standing head height, then you can either provide buttons to actually move up or down, or lean more into the game aspect and allow the user to jump. Right now it feels like you are floating around with a little dr…

Yes. Thanks for the long feedback. The controls keys are not perfect yes. WASD to add to arrows ? Yes if not to hard too dev.

As for the head, yes you are right : I should add 'something' that tilts a bit the head up/down when needed.

Re: Ask HN: What problem are you close to solving and how can we help?

#215

Post by slothtrop »

dbancajas wrote: post #129

How to find motivation/energy to do a long-term creative project when having a full time jobs + other responsibilities?

My current strat is to channel my scant motivation into maximizing my sleep and well-being, expecting to squeeze out more motivation from that.

Re: Ask HN: What problem are you close to solving and how can we help?

#216

Post by lonk11 »

dmje wrote: post #164
lonk11 wrote: post #121

Earlier quoted context omitted.

I am building a project ( https://linklonk.com ) that does information discovery in a way that respects your attention. In short, when you upvote content you connect to other people upvoted that content and to feeds that posted that content. So to get your attention other users need to prove to be a good curator of content for you. I'm planning to do a "Show HN" post next week for it and would appreciate any feedback…

Really like this idea, something I've been thinking about for a while. Will join and give the tyres a kick :-)

Thanks! It looks like 13 people signed up, which is really encouraging.

I wrote about the performance tuning I did in preparation to the Shown HN post: https://linklonk.com/item/277645707356438528

If you have any feedback please add a comment to that post.

Re: Ask HN: What problem are you close to solving and how can we help?

#217

Post by needz »

ad404b8a372f2b9 wrote: post #211
needz wrote: post #203

Earlier quoted context omitted.

The preprocessing suggestions I see are to crop out everything except for the numbers and I don’t know how to do that programmatically. There’s many kinds of displays: rollers, 7-segment, dot matrix, and LCD. The preprocessing to increase DPI to 300 did not help when I tried Tesseract, unfortunately. It’s hard to achieve a good contrast between the numbers and the backdrop

There are a lot of other options and preprocessing methods you can use to get better results. It's hard to tell without seeing the picture but thresholding/binarization might help with the contrast. In order to isolate the text, the mode option also makes a lot of difference: https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html#... If that doesn't work you'll have to add a text localization model to your pipel…

Thank you for your guidance. I will investigate further

Re: Ask HN: What problem are you close to solving and how can we help?

#218

Post by pgroves »

How to make png encoding much faster? I'm working with large medical images and after a bit of work we can do all the needed processing in under a second (numpy/scipy methods). But then the encoding to png is taking 9-15secs. As a result we have to pre-render all possible configurations and put them on S3 b/c we can't do the processing on demand in a web request.

Is there a way to use multiple threads or GPU to encode pngs? I haven't been able to find anything. The images are 3500x3500px and compress from roughly 50mb to 15mb with maximum compression (so don't say to use lower compression).

Re: Ask HN: What problem are you close to solving and how can we help?

#219

Post by graphcolorer »

I have a graph with weighted edges. I want to remove edges to make the graph colorable with N colors (e.g. N=40) such that the total weight of removed edges is minimized. If I'm able to solve this problem, that will complete a project I've been working on for years now to make a working keyboard for a person I know that has cerebral palsy.

Re: Ask HN: What problem are you close to solving and how can we help?

#220

Post by primitivesuave »

pgroves wrote: post #218

How to make png encoding much faster? I'm working with large medical images and after a bit of work we can do all the needed processing in under a second (numpy/scipy methods). But then the encoding to png is taking 9-15secs. As a result we have to pre-render all possible configurations and put them on S3 b/c we can't do the processing on demand in a web request. Is there a way to use multiple threads or GPU to encod…

I would also be interested in knowing the answer to this. Currently we use OpenSeadragon to generate a map tiling of whole slide images (~4 GB per image), then stitch together and crop tiles of a particular zoom layer to produce PNGs of the desired resolution.
Post reply on HN