The last butterfly
Midjourney 4 AI Art
#stablediffusion #dalle2 #dalle #aiart #nightcafe #midjourney

And that’s all for another day of paper reading — 7 papers boosted (out of a total of 91 today) in the cs.CV category on arXiv.org.

It feels as if I spent way more time on the 7 papers than I do on like 15 papers on other days, but that’s how it goes …

#AI #CV #NewPapers #DeepLearning #MachineLearning

"ImaginE: An Imagination-Based Automatic Evaluation Metric for Natural Language Generation. (arXiv:2106.05970v2 [cs.CL] UPDATED)" — An imagination-based automatic evaluation metric for natural language generation, which with the help of StableDiffusion, automatically generates an image as the embodied imagination for a text snippet and computes the imagination similarity using contextual embeddings.

Paper: http://arxiv.org/abs/2106.05970

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
An evaluation example on GigaWo…

This picture was accidentally taken with my iPhone 13 Pro. Being out in the country I wanted to take advantage of the lesser light pollution and get a clear image of the stars—only to get photobombed by the Milky Way. #photography

"ShiftDDPMs: Exploring Conditional Diffusion Models by Shifting Diffusion Trajectories. (arXiv:2302.02373v1 [cs.CV])" — A novel and flexible conditional diffusion model by introducing conditions into the forward process. The process utilizes extra latent space to allocate an exclusive diffusion trajectory for each condition based on some shifting rules, which will disperse condition modeling to all timesteps and improve the learning capacity of model.

Paper: http://arxiv.org/abs/2302.02373

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Exploration of the mechanism of…

@chrisvasselli Possible, these days (especially with SwiftUI) I’ve found that the error message isn’t always indicative of the actual issue 😛

@RamenCatholic We all think our thoughts are boring and everybody else has really interesting thoughts 😛 But I think what we (or at least some of us) really like is being able to share and to listen and perhaps have a conversation?

I think those are the people who thrive here because on Twitter it was all just noise. At least, in my experience …

"ReDi: Efficient Learning-Free Diffusion Inference via Trajectory Retrieval. (arXiv:2302.02285v1 [cs.CV])" — A simple yet learning-free Retrieval-based Diffusion sampling framework capable of fast inference, which retrieves a trajectory similar to the partially generated trajectory from a precomputed knowledge base at an early stage of generation, skips a large portion of intermediate steps, and continues sampling from a later step in the retrieved trajectory.

Paper: http://arxiv.org/abs/2302.02285

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Diffusion Inference (upper) and…

"Design Booster: A Text-Guided Diffusion Model for Image Translation with Spatial Layout Preservation. (arXiv:2302.02284v1 [cs.CV])" — A new approach for flexible image translation by learning a layout-aware image condition together with a text condition, which co-encodes images and text into a new domain during the training phase.

Paper: http://arxiv.org/abs/2302.02284

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Style image translation and sem…

"Divide and Compose with Score Based Generative Models. (arXiv:2302.02272v1 [cs.CV])" — Learning image components in an unsupervised manner in order to compose those components to generate and manipulate images in an informed manner.

Paper: http://arxiv.org/abs/2302.02272
Code: https://github.com/sandeshgh/Score-based-disentanglement

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Using all score components resu…

Sometimes the light just has a vibe

#photography #winter

#introduction! I'm a pro #guitarist with the band Maggie Speaks. Freelance video editor (indie doc Cover 21: Ignite Chicago, on Amazon). Amateur #photographer.

Interests:
#guitar
#music
#musician
#audio
#recording
#MusicProduction
#photography
#VideoEditing
#tech
#ios
#apple
#privacy
#history
#micromobility
#infosec
#cybersec
#law
#AI
#robotics
#ukraine
#china
#aerospace
#biotech
#UnitedKingdom
#LatAm
#EU
#austria

Languages: #english #spanish #german #espanol #deutsch #francais #italiano

@samvarma Thank you for the tip, much appreciated 🙂Engadget is probably the one account I miss over here and hopefully, this will fix that. Will try it out and thank you @vincent 🙂

@RebeccaSlatkin

"Real-Time Image Demoireing on Mobile Devices. (arXiv:2302.02184v1 [cs.CV])" — A study on accelerating demoireing networks and a dynamic demoireing acceleration method (DDA) capable of real-time deployment on mobile devices.

Paper: http://arxiv.org/abs/2302.02184
Code: https://github.com/zyxxmu/DDA

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Images with moir ́e patterns. Th…

@samvarma Thank you 🙂 Didn’t know about either one but followed now. It’s good to have some tech news … But yeah, lots of communities are harder to find on here but I’m hopeful that in a couple of years, things will improve.

But then again, when I first joined Mastodon back in 2017 or 2018, things were even more sparse and I don’t think I lasted more than a week. This time around though, things seem different …

@RebeccaSlatkin

@RamenCatholic Not sure it’s really brave, I mean on my part 😛 Just that I sometimes feel the need to write and the only way to sustain writing, at least for me, is to have even the tiniest hope that somebody reads it. So I just put it out there and keep writing. Otherwise, I’d just be writing for myself and then there’s no need to write since it’s all in my head 😃

"DeepAstroUDA: Semi-Supervised Universal Domain Adaptation for Cross-Survey Galaxy Morphology Classification and Anomaly Detection. (arXiv:2302.02005v1 [astro-ph.GA])" — A universal domain adaptation method to overcome the challenge of artificial intelligence methods extracting dataset-specific, non-robust features.

Paper: http://arxiv.org/abs/2302.02005

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Example images from the source …

@RebeccaSlatkin Mastodon really isn’t a Twitter replacement — not in the sense of being able to get people’s opinions from a large sample of people. I’ve come to accept it.

I used to follow a bunch of tech news sites so that I can get news about breaking tech news on Twitter. None of those accounts are here and I haven’t been able to find replacements. So I generally go without tech news now. Not ideal, but you do what works for you, I guess 🙂

Sure, I could go back to Twitter, but I don’t want to. I like it here. So I’ve decided to adapt to the new environment … Sure would love to find some tech news sources on here though …

@harshil@mastodon.socia

@RebeccaSlatkin @Ronnie What Apple event?

@RebeccaSlatkin A matter of perspective and usage, I think 🙂 I’ve not needed to use search once since switching to Mastodon, but then there are certainly people who depend on search a lot.

My understanding is, as @harshil said, historically people on Mastodon are opposed to text search because it can be used to harass and intimidate people. So hashtags usage is encouraged so that things can be found. But that doesn’t always work well since not everybody uses hashtags.

The other suggestion is to favourite/bookmark since you can do full text search on items you’ve favourited/bookmarked (at least on some flavours of Fedi servers). But again, not ideal since you might want to search for things that you haven’t seen.

But then again, I can also see the point of people who don’t want to be searched/found. This is probably going to get worse as more people move over to Mastodon both in terms of asking for full text search and in terms of the possibility of using search offensively increasing …