MusicLM is pretty dang cool! Even with it being kind of only a matter of time.

The fact it can create music for painting is pretty crazy. Some curveballs in there, but the model seemed to have handled it well. Wonder how cherrypicked these were.

https://google-research.github.io/seanet/musiclm/examples/

@hobs Yep, most of my posts tend to be either boost, or papers for the day that I found interesting 🙂 I do warn people about this in my profile page but most people just see one or two papers I’ve posted about and assume that they were my papers …

@hobs Don’t know if you want all the details (and this kind of evolved over time) but I originally created a system which would get the RSS feed for the cs.CV category on arXiv and use ML to summarize it for me.

But I didn’t like the summaries since they didn’t really tell me much 😛 So, I modified the system to download the PDF papers and to give me an overview of the papers each day. From that, I went to adding a form to the paper view so that I can post a summary and a page screen grab to Mastodon.

The posting system kind of fell by the wayside (I post manually now) due to various issues, but I also added a system which tracks the papers I skip and the ones I post about so that I can track things. That’s where it stands since I now post daily about new papers in the cs.CV category since some people seem to find that useful 🙂

16 papers posted out of 55 total in the cs.CV category on arXiv.org and that concludes the reading of the papers for another week!

See y'all next week 🙂

#AI #CV #NewPapers #DeepLearning #MachineLearning

"Animating Still Images. (arXiv:2209.10497v2 [cs.CV] UPDATED)" — A method for imparting motion to a still 2D image which uses deep learning to segment part of the image as the subject, uses in-paining to complete the background, and then adds animation to the subject.

Paper: http://arxiv.org/abs/2209.10497

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Interactive segmentation: Green…

"VAuLT: Augmenting the Vision-and-Language Transformer for Sentiment Classification on Social Media. (arXiv:2208.09021v3 [cs.CV] UPDATED)" — An extension of the popular Vision-and-Language Transformer (ViLT) to improve performance on vision-and-language (VL) tasks that involve more complex text inputs than image captions while having minimal impact on training and inference efficiency.

Paper: http://arxiv.org/abs/2208.09021
Code: https://github.com/gchochla/vault

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
VAuLT propagates representation…

"SIViDet: Salient Image for Efficient Weaponized Violence Detection. (arXiv:2207.12850v4 [cs.CV] UPDATED)" — A new dataset that contains videos depicting weaponized violence, non-weaponized violence, and non-violent events; and a proposal for a novel data-centric method that arranges video frames into salient images while minimizing information loss for comfortable inference by SOTA image classifiers.

Paper: http://arxiv.org/abs/2207.12850
Code: https://github.com/Ti-Oluwanimi/Violence_Detection

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Salient Image: A sequence of vi…

"BridgeTower: Building Bridges Between Encoders in Vision-Language Representation Learning. (arXiv:2206.08657v3 [cs.CV] UPDATED)" — A proposal for multiple bridge layers that build a connection between the top layers of uni-modal encoders and each layer of the cross-modal encoder.

Paper: http://arxiv.org/abs/2206.08657
Code: https://github.com/microsoft/BridgeTower

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
(a) – (d) are four categories o…

"Text-To-4D Dynamic Scene Generation. (arXiv:2301.11280v1 [cs.CV])" — A method for generating three-dimensional dynamic scenes from text descriptions which uses a 4D dynamic Neural Radiance Field (NeRF), which is optimized for scene appearance, density, and motion consistency by querying a Text-to-Video (T2V) diffusion-based model.

Paper: http://arxiv.org/abs/2301.11280

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Samples generated by MAV3D alon…

"BiBench: Benchmarking and Analyzing Network Binarization. (arXiv:2301.11233v1 [cs.CV])" — A rigorously designed benchmark with in-depth analysis for network binarization where they scrutinize the requirements of binarization in the actual production and define evaluation tracks and metrics for a comprehensive investigation.

Paper: http://arxiv.org/abs/2301.11233

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Evaluation tracks of BiBench. O…

"Improving Statistical Fidelity for Neural Image Compression with Implicit Local Likelihood Models. (arXiv:2301.11189v1 [eess.IV])" — A non-binary discriminator that is conditioned on quantized local image representations obtained via VQ-VAE autoencoders, for lossy image compression.

Paper: http://arxiv.org/abs/2301.11189

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Comparison of distortion vs. st…

"Revisiting Temporal Modeling for CLIP-based Image-to-Video Knowledge Transferring. (arXiv:2301.11116v1 [cs.CV])" — A look at temporal modeling in the context of image-to-video knowledge transferring, which is the key point for extending image-text pretrained models to the video domain.

Paper: http://arxiv.org/abs/2301.11116

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
(a) Illustration of temporal mo…

"Explaining Visual Biases as Words by Generating Captions. (arXiv:2301.11104v1 [cs.LG])" — Diagnosing the potential biases in image classifiers by leveraging two types (generative and discriminative) of pre-trained vision-language models to describe the visual bias as a word.

Paper: http://arxiv.org/abs/2301.11104

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Concept of the proposed bias-to…

"Vision-Language Models Performing Zero-Shot Tasks Exhibit Gender-based Disparities. (arXiv:2301.11100v1 [cs.CV])" — An exploration of the extent to which zero-shot vision-language models exhibit gender bias for different vision tasks.

Paper: http://arxiv.org/abs/2301.11100

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
(a) The average precision (AP) …

"simple diffusion: End-to-end diffusion for high resolution images. (arXiv:2301.11093v1 [cs.CV])" — Improve denoising diffusion for high resolution images while keeping the model as simple as possible and obtaining performance comparable to the latent diffusion-based approaches?

Paper: http://arxiv.org/abs/2301.11093

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
A dslr photo of a frog wearing…

Another of my fluid abstractions. 'Awestruck.' Think I should just call them my restlessness series since i only draw them when I'm stuck wanting to do 5 things at once and not being able to prioritize. #mastoart #abstract #illustration

I went to two exhibits this week; one by Sidne Teske, and another by Phyllis Shafer; my second visit before the show closes tomorrow. So fortunate to have opportunities to see amazing artwork in person. Always make me want to hurry home and get to work in my studio. Artwork is hard work.

#MastoArt #fediart #womensart #artistsonmastodon

"Rewarded meta-pruning: Meta Learning with Rewards for Channel Pruning. (arXiv:2301.11063v1 [cs.CV])" — A method to reduce the parameters and FLOPs for computational efficiency in deep learning models by introducing accuracy and efficiency coefficients to control the trade-off between the accuracy of the network and its computing efficiency.

Paper: http://arxiv.org/abs/2301.11063

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Reward at varying Accuracies an…

"Explore the Power of Dropout on Few-shot Learning. (arXiv:2301.11015v1 [cs.CV])" — An exploration of the power of the dropout regularization technique on few-shot learning and provide some insights about how to use it.

Paper: http://arxiv.org/abs/2301.11015

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
The generalization power of tra…

"On the Importance of Noise Scheduling for Diffusion Models. (arXiv:2301.10972v1 [cs.CV])" — A study of the effect of noise scheduling strategies for denoising diffusion generative models which finds that the noise scheduling is crucial for performance.

Paper: http://arxiv.org/abs/2301.10972

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Random samples generated by our…