This Instant And Moment - 2025!

Putting a claim in to claw back some of the cost of my root canal and upcoming crown. I thought I could claim back £450 for each treatment, I can't. It's £450 maximum per year for restorative treatments. :(

Shame, especially given how close you are to spanning 2 years!
 
Shame, especially given how close you are to spanning 2 years!
I know! I was trying to push it out to the last week in March and then the crown in April to claim twice m, but it’s been going on since Sept and I didn’t want to risk it getting infected again :(
 
Amazon knows me well.

cI9d8fM.jpeg
 
So tired, so nearly home time... Just have to shop, hoover, empty bins and cat litter and clean the bathroom, then we're ready for family to visit tomorrow morning!
 
It just occurred to me all the data entry jobs that AI killed off got replaced by image and video annotation for machine learning datasets, which is possibly even more tedious.

wonder how much it pays at a good company, or do all the big tech companies outsource?
 
It just occurred to me all the data entry jobs that AI killed off got replaced by image and video annotation for machine learning datasets, which is possibly even more tedious.

wonder how much it pays at a good company, or do all the big tech companies outsource?

There are pretty decently paid jobs in AI training, because they need people who are knowledgeable and dilligent to do it. I get adverts about it from time to time but I'm no scab.
 
There are pretty decently paid jobs in AI training, because they need people who are knowledgeable and dilligent to do it. I get adverts about it from time to time but I'm no scab.
I'm taking about the people who draw a box around an object in a video or image and give it a label, move on to the next frame, ensure the auto tracking moved the boxes into the right areas, next frame.
They don't need to know any code,. they don't actually train the model which can mostly be automated anyway. but AFAIK there's no short cuts for annotation, only crowd sourced annotators

surely they don't get paid much? although I guess they have to be paid enough to do the boxes properly in the first place

Imagine how much work it would be if you wanted to annotate a 90 minute movie it must take a crazy amount of man hours??


chatgpt tells me theres no secret workarounds to speed everything up apart from the boxes auto tracking over frames...

there's no way to automate it with computers, because they have to already have been trained and have the data.

people need to annotate so they can train etc


even like a 6 second video at 60fps with around 10 moving elements on the screen is crazy annoying to annotate

have to move every box to be properly centere in most frames.
have to hide the label/bounding box on any frames they aren't visible.
make the box larger or smaller for frames as needed.
check when new things appear and things leave for good.

you;ll see videos online of CVAT tracking bounding boxes like magic, when you actually use their website it's nothing like that at all. the boxes move approximately in the right place but still need adjustment.



if anyone wants a bird feeder these look sick for 9.99
weirdly was like 18£ when I ordered, but amazon refunded me the difference.


all metal apart from the plastic perch and the plastic bottom.
it does come with 2 spare perch incase they get broken them when taking it apart to clean.
it's painted metal.

they are kinda wide compared to most tube feeders, I was kinda surprised at the size.

looks nice though

NOT MY VIDEO

only just occurred to me recently that bird feeders are basically emulating how birds would feed naturally at a dead tree looking in holes for insects etc..

yet people claim birds became reliant on them.. surely the bird thinks it's foraging as normal/
 
Last edited:
I feel like this is not the kind of answer you should assume chatGPT is right about... Concerning.
no i looked into it on google, it seems where's barely any software for annotation and it's all basically all manual.

unless you already annotated a dataset for a model to learn on in the first place, then it can annotate the same stuff fine.

I tried annotating by template matching on a video but even with a 4090 it was crazy slow like 0.1frames a second so I gave up on that..

seems it takes just way too much processing power if you have a lot of objects, I had about 60+ some of them having multiple images per object (front, left, right, back sides etc or different stances)


I tried with synthetic data too but it doesn't work right and learns only to detect synthetic data :S , maybe I should test it again but have the bounding boxes shift around by a fix pixels randomly in each image...

The best way for a simple 2d game seems to be only record tiny videos of the different mobs/pick-ups etc
then manually annotate, even like 6 seconds is more than enough frames of data.

it's just annoying.
 
Last edited:
Just had a horrendous drive home in the weather conditions - one place was flooded right across the road at least a foot deep probably more like 18 inches for 50 yards or so - even having some idea how to drive through water it was sketchy and I wouldn't have done it in hindsight - normally even when it is this bad the highest point of the road isn't more than 6 inches deep, not this time.
 
New cat has been missing since Friday night :(

She's only around 6 months and only been going out for around a month but has already seemed to be more adventurous than our other cat who rarely ventures far, so not really sure how far away she might be. Also the concern there's a lot of outdoor dogs around where we live (proper aggresive outdoor guard dogs) so always a risk that one of those has got her and if not one of them then there are also lots of wolves/foxes around.

We've been out on a lot of the local roads to see if she's been hit by a car but no sign there and we're away from most main roads so people don't tend to drive too fast so it's less of a risk.
 
I wasnt sure after yesterdays torrential wind and rain that i would make the cycle in to work today, but my bike got some use this morning,

Hopefully it doesnt pelt down with rain later before i leave

Incentive ? im not going to turn down free food.
 
Last edited:
I wasnt sure after yesterdays torrential wind and rain that i would make the cycle in to work today, but my bike got some use this morning,

Hopefully it doesnt pelt down with rain later before i leave

Incentive ? im not going to turn down free food.
I just dropped my girl at school and managed to avoid a torrential downpour. Fingers crossed for 3pm :D
 
Back
Top Bottom