Large vision-language models (VLMs) commonly process images at native or high resolution to remain effective across tasks.This inflates visual tokens up to to 99% of total tokens of the prefill stage, resulting in high compute and latency, even when lowresolution images would suffice.We introduce CARES-a Context-Aware Resolution Selector, a lightweight preprocessing module that, given an image-query pair, predicts the minimal sufficient input resolution.CARES uses a compact VLM (350M) to extract features and predict when a target pretrained VLM's response converges to its peak ability to answer correctly.Though trained as a discrete classifier over a set of optional resolutions, CARES interpolates continuous resolutions at inference for fine-grained control.Across nine multimodal benchmarks spanning documents and natural images, as well as diverse target VLMs, CARES preserves task performance while reducing compute by up to 78% on average across 9 benchmarks.

Image reconstruction from corrupted images is crucial across many domains. Most reconstruction networks are trained on post-ISP sRGB images, even though the image-signal-processing pipeline irreversibly mixes colors, clips dynamic range and blurs fine detail. This paper uses the rain degradation problem as a "use case" to show that these losses are avoidable and show that learning directly on raw Bayer mosaics yields superior reconstructions. To substantiate the claim we (i) evaluate post-ISP and Bayer reconstruction pipelines, (ii) curate Raw-Rain, the first public benchmark of real rainy scenes captured in both 12-bit Bayer and bit-depth-matched sRGB, and (iii) introduce Information Conservation Score (ICS), a color-invariant metric that aligns more closely with human opinion than PSNR or SSIM. On the test split our raw-domain model improves sRGB results by up to +0.99 dB PSNR and +1.2 % ICS, while running faster with lower GFLOPs. The results advocate an ISP-last paradigm for low-level vision and open the door to end-to-end learnable camera pipelines.

Pre-trained Vision Transformers now serve as powerful tools for computer vision. Yet, efficiently adapting them for multiple tasks remains a challenge that arises from the need to modify the rich hidden representations encoded by the learned weight matrices, without inducing interference between tasks. Current parameter-efficient methods like LoRA, which apply low-rank updates, force tasks to compete within constrained subspaces, ultimately degrading performance. We introduce DiTASK a novel Diffeomorphic Multi-Task Fine-Tuning approach that maintains pre-trained representations by preserving weight matrix singular vectors, while enabling task-specific adaptations through neural diffeomorphic transformations of the singular values. By following this approach, DiTASK enables both shared and task-specific feature modulations with minimal added parameters. Our theoretical analysis shows that DiTASK achieves full-rank updates during optimization, preserving the geometric structure of pretrained features, and establishing a new paradigm for efficient multi-task learning (MTL). Our experiments on PASCAL MTL and NYUD show that DiTASK achieves state-of-the-art performance across four dense prediction tasks, using 75% fewer parameters than existing methods. Our code is available here.

Robust classification methods predominantly concen-trate on algorithms that address a specific threat model, resulting in ineffective defenses against other threat models. Real-world applications are exposed to this vulnerability, as malicious attackers might exploit alternative threat models. In this work, we propose a novel test-time threat model agnostic algorithm that enhances Adversarial-Trained (AT) models. Our method operates through COnditional image transformation and DIstance-based Prediction (CODIP) and includes two main steps: First, we transform the input image into each dataset class, where the input image might be either clean or attacked. Next, we make a prediction based on the shortest transformed distance. The conditional transformation utilizes the perceptually aligned gradients property possessed by AT models and, as a result, elimi-nates the need for additional models or additional training. Moreover, it allows users to choose the desired balance between clean and robust accuracy without training. The proposed method achieves state-of-the-art results demon-strated through extensive experiments on various models, AT methods, datasets, and attack types. Notably, applying CODIP leads to substantial robust accuracy improvement of up to$+23\%, +20\%, +26\%$, and +22% on CIFAR10, CIFAR100, ImageNet and Flowers datasets, respectively. For more details, visit the project page.

Concept bottleneck models (CBMs) have emerged as critical tools in domains where interpretability is paramount. These models rely on predefined textual descriptions, referred to as concepts, to inform their decision-making process and offer more accurate reasoning. As a result, the selection of concepts used in the model is of utmost significance. This study proposes Conceptual Learning via Embedding Approximations for Reinforcing Interpretability and Transparency, abbreviated as CLEAR, a framework for constructing a CBM for image classification. Using score matching and Langevin sampling, we approximate the embedding of concepts within the latent space of a vision-language model (VLM) by learning the scores associated with the joint distribution of images and concepts. A concept selection process is then employed to optimize the similarity between the learned embeddings and the predefined ones. The derived bottle-neck offers insights into the CBM's decision-making process, enabling more comprehensive interpretations. Our approach was evaluated through extensive experiments and achieved state-of-the-art performance on various benchmarks. The code for our experiments is available at https://github.com/clearProject/CLEAR/tree/main.

The ability of robots to manipulate objects relies heavily on their aptitude for visual perception. In domains charac-terized by cluttered scenes and high object variability such as traffic, navigation and object grasping, most methods call for vast labeled datasets, laboriously hand-annotated, with the aim of training capable models. Once deployed, the challenge of generalizing to unfamiliar objects implies that the model must evolve alongside its domain. To address this, we propose a novel framework that combines Semi-Supervised Learning (SSL) with Learning Through Interaction (LTI), allowing a model to learn by observing scene alterations and leverage visual consistency despite tempo-ral gaps without requiring curated data of interaction se-quences. As a result, our approach exploits partially anno-tated data through self-supervision and incorporates temporal context using pseudo-sequences generated from unla-beled still images. We validate our method on two common benchmarks, ARMBench mix-object-tote and OCID, where it achieves state-of-the-art performance. Notably, on ARM-Bench, we attain an AP50 of 86.37, almost a 20% improvement over existing work, and obtain remarkable results in scenarios with extremely low annotation, achieving an AP50score of 84.89 with just 1 % of annotated data compared to previous state of the art of 82 which targeted the fully anno-tated dataset.

Multi-rotor aerial autonomous vehicles (MAVs) primarily rely on vision for navigation purposes. However, visual localization and odometry techniques suffer from poor performance in low or direct sunlight, a limited field of view, and vulnerability to occlusions. Acoustic sensing can serve as a complementary or even alternative modality for vision in many situations, and it also has the added benefits of lower system cost and energy footprint, which is especially important for micro aircraft. This paper proposes actively controlling and shaping the aircraft propulsion noise generated by the rotors to benefit localization tasks, rather than considering it a harmful nuisance. We present a neural network architecture for self-noise-based localization in a known environment. We show that training it simultaneously with learning time-varying rotor phase modulation achieves accurate and robust localization. The proposed methods are evaluated using a computationally affordable simulation of MAV rotor noise in 2D acoustic environments that is fitted to real recordings of rotor pressure fields. Code3and data4are accompanied.

Message Passing Graph Neural Networks (MPGNNs) have emerged as the preferred method for modeling complex interactions across diverse graph entities. While the theory of such models is well understood, their aggregation module has not received sufficient attention. Sum-based aggregators have solid theoretical foundations regarding their separation capabilities. However, practitioners often prefer using more complex aggregations and mixtures of diverse aggregations. In this work, we unveil a possible explanation for this gap. We claim that sum-based aggregators fail to "mix" features belonging to distinct neighbors, preventing them from succeeding at downstream tasks. To this end, we introduce Sequential Signal Mixing Aggregation (SSMA), a novel plug-and-play aggregation for MPGNNs. SSMA treats the neighbor features as 2D discrete signals and sequentially convolves them, inherently enhancing the ability to mix features attributed to distinct neighbors. By performing extensive experiments, we show that when combining SSMA with well-established MPGNN architectures, we achieve substantial performance gains across various benchmarks, achieving new state-of-the-art results in many settings. We published our code at https://almogdavid.github.io/SSMA/.

Traditional methods for learning with the presence of noisy labels have successfully handled datasets with artificially injected noise but still fall short of adequately handling real-world noise. With the increasing use of meta-learning in the diverse fields of machine learning, researchers leveraged auxiliary small clean datasets to meta-correct the training labels. Nonetheless, existing meta-label correction approaches are not fully exploiting their potential. In this study, we propose an Enhanced Meta Label Correction approach abbreviated as EMLC for the learning with noisy labels (LNL) problem. We re-examine the meta-learning process and introduce faster and more accurate meta-gradient derivations. We propose a novel teacher architecture tailored explicitly to the LNL problem, equipped with novel training objectives. EMLC outperforms prior approaches and achieves state-of-the-art results in all standard benchmarks. Notably, EMLC enhances the previous art on the noisy real-world dataset Clothing1M by 1.52% while requiring ×0.5 the time per epoch and with much faster convergence of the meta-objective when compared to the baseline approach.1

The referring video object segmentation task (RVOS) involves segmentation of a text-referred object instance in the frames of a given video. Due to the complex nature of this multimodal task, which combines text reasoning, video understanding, instance segmentation and tracking, existing approaches typically rely on sophisticated pipelines in order to tackle it. In this paper, we propose a simple Transformer-based approach to RVOS. Our framework, termed Multimodal Tracking Transformer (MTTR), models the RVOS task as a sequence prediction problem. Following recent advancements in computer vision and natural language processing, MTTR is based on the realization that video and text can be processed together effectively and elegantly by a single multimodal Transformer model. MTTR is end-to-end trainable, free of text-related inductive bias components and requires no additional mask-refinement post-processing steps. As such, it simplifies the RVOS pipeline considerably compared to existing methods. Evaluation on standard benchmarks reveals that MTTR significantly outperforms previous art across multiple metrics. In particular, MTTR shows impressive +5.7 and +5.0 mAP gains on the A2D-Sentences and JHMDB-Sentences datasets respectively, while processing 76 frames per second. In addition, we report strong results on the public validation set of Refer-YouTube-VOS, a more challenging RVOS dataset that has yet to receive the attention of researchers. The code to reproduce our experiments is avail-able at https://github.com/mttr2021/MTTR.

The success of learning with noisy labels (LNL) methods relies heavily on the success of a warm-up stage where standard supervised training is performed using the full (noisy) training set. In this paper, we identify a "warm-up obstacle": the inability of standard warm-up stages to train high quality feature extractors and avert memorization of noisy labels. We propose "Contrast to Divide" (C2D), a simple framework that solves this problem by pre-training the feature extractor in a self-supervised fashion. Using self-supervised pre-training boosts the performance of existing LNL approaches by drastically reducing the warm-up stage's susceptibility to noise level, shortening its duration, and improving extracted feature quality. C2D works out of the box with existing methods and demonstrates markedly improved performance, especially in the high noise regime, where we get a boost of more than 27% for CIFAR-100 with 90% noise over the previous state of the art. In real-life noise settings, C2D trained on mini-WebVision outperforms previous works both in WebVision and ImageNet validation sets by 3% top-1 accuracy. We perform an in-depth analysis of the framework, including investigating the performance of different pre-training approaches and estimating the effective upper bound of the LNL performance with semi-supervised learning. Code for reproducing our experiments is available at https://github.com/ContrastToDivide/C2D.

One major obstacle hindering the ubiquitous use of CNNs for inference is their relatively high memory bandwidth requirements, which can be the primary energy consumer and throughput bottleneck in hardware accelerators. Inspired by quantization-aware training approaches, we propose a compression-aware training (CAT) method that involves training the model to allow better compression of weights and feature maps during neural network deployment. Our method trains the model to achieve low-entropy feature maps, enabling efficient compression at inference time using classical transform coding methods. CAT significantly improves the state-of-the-art results reported for quantization evaluated on various vision and NLP tasks, such as image classification (ImageNet), image detection (Pascal VOC), sentiment analysis (CoLa), and textual entailment (MNLI). For example, on ResNet-18, we achieve near baseline ImageNet accuracy with an average representation of only 1.5 bits per value with 5-bit quantization. Moreover, we show that entropy reduction of weights and activations can be applied together, further improving bandwidth reduction. Reference implementation is available.

Neural network quantization enables the deployment of large models on resource-constrained devices. Current post-training quantization methods fall short in terms of accuracy for INT4 (or lower) but provide reasonable accuracy for INT8 (or above). In this work, we study the effect of quantization on the structure of the loss landscape. We show that the structure is flat and separable for mild quantization, enabling straightforward post-training quantization methods to achieve good results. We show that with more aggressive quantization, the loss landscape becomes highly non-separable with steep curvature, making the selection of quantization parameters more challenging. Armed with this understanding, we design a method that quantizes the layer parameters jointly, enabling significant accuracy improvement over current post-training quantization methods. Reference implementation is available at https://github.com/ynahshan/nn-quantization-pytorch/tree/master/lapq .

Convolutional neural networks (CNNs) achieve state-of-the-art accuracy in a variety of tasks in computer vision and beyond. One of the major obstacles hindering the ubiquitous use of CNNs for inference on low-power edge devices is their high computational complexity and memory bandwidth requirements. The latter often dominates the energy footprint on modern hardware. In this paper, we introduce a lossy transform coding approach, inspired by image and video compression, designed to reduce the memory bandwidth due to the storage of intermediate activation calculation results. Our method does not require fine-tuning the network weights and halves the data transfer volumes to the main memory by compressing feature maps, which are highly correlated, with variable length coding. Our method outperform previous approach in term of the number of bits per value with minor accuracy degradation on ResNet-34 and MobileNetV2. We analyze the performance of our approach on a variety of CNN architectures and demonstrate that FPGA implementation of ResNet-18 with our approach results in a reduction of around 40% in the memory energy footprint, compared to quantized network, with negligible impact on accuracy. When allowing accuracy degradation of up to 2%, the reduction of 60% is achieved. A reference implementation accompanies the paper.

"Beauty is in the eye of the beholder." This maxim, emphasizing the subjectivity of the perception of beauty, has enjoyed a wide consensus since ancient times. In the digital era, data-driven methods have been shown to be able to predict human-assigned beauty scores for facial images. In this work, we augment this ability and train a generative model that generates faces conditioned on a requested beauty score. In addition, we show how this trained generator can be used to "beautify" an input face image. By doing so, we achieve an unsupervised beautification model, in the sense that it relies on no ground truth target images. Our implementation is available on: https://github.com/beholdergan/Beholder-GAN.

We present a novel method for neural network quantization. Our method, named UNIQ , emulates a non-uniform k -quantile quantizer and adapts the model to perform well with quantized weights by injecting noise to the weights at training time. As a by-product of injecting noise to weights, we find that activations can also be quantized to as low as 8-bit with only a minor accuracy degradation. Our non-uniform quantization approach provides a novel alternative to the existing uniform quantization techniques for neural networks. We further propose a novel complexity metric of number of bit operations performed (BOPs), and we show that this metric has a linear relation with logic utilization and power. We suggest evaluating the trade-off of accuracy vs. complexity (BOPs). The proposed method, when evaluated on ResNet18/34/50 and MobileNet on ImageNet, outperforms the prior state of the art both in the low-complexity regime and the high accuracy regime. We demonstrate the practical applicability of this approach, by implementing our non-uniformly quantized CNN on FPGA.