Machine Learning

1 readers

1 users here now

Community Rules:

Be nice. No offensive behavior, insults or attacks: we encourage a diverse community in which members feel safe and have a voice.
Make your post clear and comprehensive: posts that lack insight or effort will be removed. (ex: questions which are easily googled)
Beginner or career related questions go elsewhere. This community is focused in discussion of research and new projects that advance the state-of-the-art.
Limit self-promotion. Comments and posts should be first and foremost about topics of interest to ML observers and practitioners. Limited self-promotion is tolerated, but the sub is not here as merely a source for free advertisement. Such posts will be removed at the discretion of the mods.

founded 1 year ago

MODERATORS

communick@academy.garden

[D] model for "understanding" dashcam images (alien.top)

submitted 1 year ago by samrus@alien.top to c/machinelearning@academy.garden

2 comments fedilink hide all child comments

i have some dashcam footage from my car and want to see how a model could embed the images in an unsupervised (or self-supervised) way so i dont have to label everything.

like if scenarios that are semantically similar, but different in pixel space (pulling out the driveway in the day versus pulling in at night) could be clustered close-ish together in latent space so that i could label fewer images and have the model get the other using something like k-nearest or whatever.

i am starting off with just frame level before i try to tackle videos as a sequence of images (will probably lose interest by that point, so want to get images dealt with first). i looked in to VAEs and tried training one from scratch on my data but i dont have enough compute power for that.

does anyone here have any ideas about this? any pretrained off the shelf models that i could use for this? any leads for a literature survey?

you are viewing a single comment's thread
view the rest of the comments

[–] seiqooq@alien.top 1 points 1 year ago

I’ve personally used SimCLR with some success but unless you doctor the embedding scheme, similarity will primarily be a function of pixel likeness