Protocol7 posted:

I think SQLAlchemy in particular has one oddity where if you try to count from a table, it does a SELECT * FROM instead of something like SELECT 1 FROM which causes some performance issues as the target table increases in size and complexity.

Protocol7 posted:

I think SQLAlchemy in particular has one oddity where if you try to count from a table, it does a SELECT * FROM instead of something like SELECT 1 FROM which causes some performance issues as the target table increases in size and complexity.

redleader posted:

this occurs when your database is an idiot

nightbae smokewheat posted:

I'm goofing around with image recognition software for a hobby project, and my end goal is analogous to car counting as demonstrated here:

https://www.youtube.com/watch?v=n_Og933KtAE

So far, I've figured out how to install labelImg and ImageAI and use these to annotate a training image dataset, and to train a YOLOv3 model that can detect car analogs in still images and video. Today I realized that to do counting I need to perform not only object detection, but object tracking as well, and I don't think that ImageAI supports the latter.

1) Do I need to drop ImageAI and cobble together something like this or this instead?

2) In case there is a go-to solution for this that I'm too stupid to have found already: can anyone recommend their favorite open source solution for object detection and tracking in video?

Computer viking posted:

Though that said, being able to flag it when cars do break out of their path, intersect, and/or release fragments could perhaps be useful on its own.

nightbae smokewheat posted:

Thanks; yes, I have seen deepsort come up in a few searches too so I will start looking into that this weekend. Fortunately I'm not a state-level actor, but instead just a garden-variety dumbass, so I don't have to worry about my car analogues dropping chaff or otherwise actively breaking track.

Volmarias posted:

Relevant:

https://www.youtube.com/watch?v=k7YVxLLIuGM

hbag posted:

cum wahoo

nightbae smokewheat posted:

1) Do I need to drop ImageAI and cobble together something like this or this instead?

2) In case there is a go-to solution for this that I'm too stupid to have found already: can anyone recommend their favorite open source solution for object detection and tracking in video?

hbag posted:

i uh
im getting this in python
AttributeError: 'dict' object has no attribute 'descendants'

this makes no sense, i thought dicts did have descendants, that was the whole point
idk heres my code lmao

Python code:

            for post in posts:
                #resp_object[post] = 'wahooo'
                resp_object['cum'].append('wahoo')

neither of these worked

nielsm posted:

hbag posted:

i uh
im getting this in python
AttributeError: 'dict' object has no attribute 'descendants'

this makes no sense, i thought dicts did have descendants, that was the whole point
idk heres my code lmao

Python code:

            for post in posts:
                #resp_object[post] = 'wahooo'
                resp_object['cum'].append('wahoo')

neither of these worked

DoctorTristan posted:

What this guy said, also - not saying this is drawn from experience or anything - but you can never really predict where a piece of code you write is going to end up or who might demand to see it, so maybe dance like nobody’s watching but pick variable names like they’ll be read out in a televised courtroom

hbag posted:

trust me im not pushing anything with "cum wahoo" in it to my git repo

anyway heres more of the code lmao

Python code:

def viewthread_actual(thread_id):
    if verifyCall():
        sesh = requests.Session()
        sesh.cookies = http.cookiejar.MozillaCookieJar("cookies.txt")
        forumAuth(sesh)

	# query thread
        if 'page' not in request.args:
            page_num = 1
        else:
            page_num = request.args.get('page')
        threadRaw = sesh.post(f'https://forums.somethingawful.com/showthread.php?threadid={thread_id}&userid=0&perpage=40&pagenumber={page_num}')
        thread = BeautifulSoup(threadRaw.content, 'html.parser')
        
        # grab thread
        if thread.select_one('.userinfo .author') != None:
            resp_object = {'bing bing': 'wahoo'}
            posts = BeautifulSoup.findAll({'class': 'post'})
            for post in posts:
                #resp_object[post].append('wahooo')
                resp_object['cum'].append('wahoo')
        else:
            resp_object = {'error': 'Thread not found.'}
    else:
        resp_object = {'error': 'Unauthorized Access. Ensure your PUBLIC token is somewhere on your profile, and you\'ve included your PRIVATE key in your request.'}
    
    return resp_object

im pretty sure it is a beautifulsoup issue which is why i commented out the line that uses poo poo from beautifulsoup and just tried to set a fixed key pair but that didnt seem to work either

pokeyman posted:

It doesn't match the error message you posted before, but I think you need to put a value in resp_object for the key 'cum' before you try to append to it?

After that, I'm not sure exactly what using BeautifulSoup elements as dict keys will do, but my guess is "end poorly". In the commented-out bit, did you mean to do resp_object[‘post’] or something?

hbag posted:

No, I meant to do resp_object[post], which I know isn't great but it's just while i get it up and running.
As for your first point, I also tried doing resp_object['foo'] = 'bar', but that had the same error.

For context, I'm trying to get a JSON object of variable length, since this function I'm writing reads a thread and has each post in its own key pair in the response object. So, naturally, it'd need to append each new post it finds to the object.

pokeyman posted:

Well I'm baffled on the descendants error. The "each post in the response" part makes total sense, but I would've expected resp_object[post.id] = post or something, i.e. using a string value as the key.

hbag posted:

yeah it MIGHT be because its trying to name the key after the entire 'post' object or some poo poo, ill try some more poo poo and see if that works

edit: actually wait no that doesnt make sense either because it has the exact same error when i just try it with a hardcoded value

pokeyman posted:

Welp. Well if it helps, I have limitless patience for anyone trying to use the SA Forums's unofficial API.

I complimented someone the other day about how good they'd gotten at picking out the problematic line from a backtrace. And I took a minute to phrase it because I was worried it would sound patronizing, but I was completely genuine. Very useful skill!

hbag posted:

oh gdi i was looking at the wrong thing
it was being thrown by line 34, which is this:

Python code:

posts = BeautifulSoup.findAll({'class': 'post'})

hbag posted:

is there another unofficial API or are you talking about the one i wrote (at https://api.fyad.club) that this is for

KillHour posted:

Well that's a lot easier to answer. First, findAll has been deprecated and moved to find_all:

DoctorTristan posted:

Is this… package maintainers actually deciding to give a poo poo about PEP8?

hbag posted:

camelcase is for functions, underscores are for variables

Gin_Rummy posted:

Can anyone ELI5 exactly what companies/job postings mean when they refer to experience working with cloud infrastructure like AWS and/or GoogleCloud?

ultrafilter posted:

Posting here cause I can't think of any better place to ask:

My team is using GitLab to host our repositories and do CI on merge requests. We want to split our CI build into different stages so that we can get useful information about what's failing, but GitLab has the possibility of copying the artifacts from any given stage to a different directory or even a different machine to do the next stage. Our artifacts are about 2GB total, so that hurts. Is there a way to get it to run the entire pipeline in a single directory? I know that Jenkins can do that, so it's not a foreign concept, but searching any of the obvious keywords doesn't help.