I need this device and was thinking about making my own. One of my cats is athletic and jumps up to the second floor balcony to get in via one of our bedroom windows. It is very easy at 3:00 AM to accidentally let the cat in while it has a still alive mouse, rat, bunny, chipmunk, etc..
If you've only got a few classes it's faster to transfer learn a model from a few hundred examples, get that model to relabel your entire dataset and then fix any mistakes it's made. This is less than ten lines of python using fast ai.
If you've split your labelled data into folders then you can scan through several thumbnails at once to catch any misclassified pictures.
This might be the most practical way to approach this, but it's not really a fundamental improvement since you still need to consider every image (also the correctly labeled ones).
True, although as the sibling suggested you can sort by confidence and weed out the edge cases. It's also quicker to go through mostly correct images as you need to make fewer (corrective) interactions/clicks. You can even view eg four thumbnails at once to spot outliers.
Not necessarily. The classifier will not only tell you whether it sees a rat, but also its confidence. So you can manually classify just the low confidence images, retrain, and repeat until there's none left or you're happy with its detection rate.
Running a computer vision consulting shop, that's our way of life. For almost a decade we've developed semi-automated means to make the problem saner, but we have to maintain a network of people and software for labeling images.
Once you have each image in its own file, not hard to rig up a way to preview them one at a time and press a number key corresponding to the desired category (1–4 in this case) which will bin the image in that category and bring up the next image.
Faster if you can do them in a batch. If you have individual video segments, just label the whole segment. There isn't going to be no prey in one frame and then suddenly prey materializes from nowhere in the next frame.
Brute force pattern matching would be cataloguing all possible matching pixel patterns. The whole point of the ML approach is that with a comparatively few images (yes even 20k is comparatively few) the computer will correctly match pixel-sets it has never encountered before.
If your comparison function returns a float instead of a binary value, then you don't need to match against every possible image. Instead you just return how much each image is "similar" to the given image. A simple approach is to use cross correlation for the similarity.
Aside: why is it so difficult to upload talks to YouTube with good audio?
The audio in this video was constantly clipping which forced me to turn my speakers way down. We live in a time where we can buy off-the-shelf hardware to do machine learning / image recognition, but can't record a presenter's speech properly?
Edit: it looks like we've had to ask you this many times before and also that you've made a habit of posting unsubstantive comments to HN. Could you please not do those things any more? We'd greatly appreciate it.
How about not letting your invasive species drive other animals to extinction for sport? It’s nonsensical to call it a natural animal, and completely ignore the fact that they live nothing like an animal in nature, and have their population artificially supported while they decimate other species for sport. A house cat shares 0% of the life of a ‘natural’ animal.
If you want to own a pet cat, don’t promote irresponsible behavior that intentionally harms the environment. Attempt to have a modicum respect for things that exist outside yourself.
> Attempt to have a modicum respect for things that exist outside yourself
Personal attacks will get you banned here, regardless of how strongly you feel about cats. Would you please review https://news.ycombinator.com/newsguidelines.html and use HN as intended when posting to it?
It’s not a binary choice between owning a cat and letting it wipe out as much of the native animal population as it can. Again, encouraging people to let their animals wipe out populations of wild animals is wrong, and doing it because it allows your domesticated animal the freedom of feeling ‘natural’ is a supremely selfish behavior.
Indoor cats live longer and don't have problems with parasites, pregnancy, diseases, fights, poisons, predators, cars, stupid humans, or just getting lost.
Or ideally just don't buy a cat at all, since they carry toxoplasmosis and their scratches are linked to depression(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4700762/). Every time a cat owner commits suicide I can't help but wonder if that could have been easily prevented.
Correlation does not imply causation and there still isn't enough evidence to support the claim that cat scratches are linked to depression.
From the abstract:
"Authors suggested that infection with cat parasite Toxoplasma could be the reason for this association."
From the conclusion:
"Absence of association between toxoplasmosis and depression and five times stronger association of depression with cat scratching than with cat biting suggests that the pathogen responsible for mood disorders in animals-injured subjects is probably not the protozoon Toxoplasma gondii but another organism; possibly the agent of cat-scratched disease – the bacteria Bartonella henselae."
>Every time a cat owner commits suicide I can't help but wonder if that could have been easily prevented.
There are close to a billion people with cats, and like a few tens of thousands suicides, that include tons of people with no cats at all.
Especially since even your article says "Cat biting and toxoplasmosis had no effect on the depression, and the number of cats at home had a negative effect on depression (p = 0.021)."
So, only cat scratching does? As if the people with many cats at home don't get scratch? I wouldn't place much emphasis on a single study of questionable quality (given ample empirical evidence that there's no big deal).
"Cats cause suicides" border on "no vaccines" territory...
