DNA storage has matured from concept to practical stage, yet its integration with neural compression pipelines remains inefficient. Early DNA encoders applied redundancy-heavy constraint layers atop raw binary data-workable but primitive. Recent neural codecs compress data into learned latent representations with rich statistical structure, yet still convert these latents to DNA via naive binary-to-quaternary transcoding, discarding the entropy model's optimization. This mismatch undermines compression efficiency and complicates the encoding stack. A plug-in module that collapses latent compression and DNA encoding into a single step. SCONE performs quaternary arithmetic coding directly on the latent space in DNA bases. Its Constraint-Aware Adaptive Coding module dynamically steers the entropy encoder's learned probability distribution to enforce biochemical constraints-GC balance and homopolymer suppression-deterministically during encoding, eliminating post-hoc correction. The design preserves full reversibility and exploits the hyperprior model's learned priors without modification. Experiments show SCONE achieves near-perfect constraint satisfaction with negligible computational overhead (< 2% latency), establishing a latent-agnostic interface for end-to-end DNA-compatible learned codecs.

Neural architecture search (NAS) has emerged as a powerful paradigm that enables researchers to automatically explore vast search spaces and discover efficient neural networks. However, NAS suffers from a critical bottleneck, i.e., the evaluation of numerous architectures during the search process demands substantial computing resources and time. In order to improve the efficiency of NAS, a series of methods have been proposed to reduce the evaluation time of neural architectures. However, they are not efficient enough and still only focus on the accuracy of architectures. Beyond classification accuracy, real-world applications increasingly demand more efficient and compact network architectures that balance multiple performance criteria. To address these challenges, we propose the SMEMNAS, a pairwise comparison relation-assisted multiobjective evolutionary algorithm (EA) based on a multipopulation (MP) mechanism. In the SMEMNAS, a surrogate model is constructed based on pairwise comparison relations to predict the accuracy ranking of architectures, rather than the absolute accuracy. Moreover, two populations cooperate with each other in the search process, i.e., a main population that guides the evolutionary process and a vice population that enhances search diversity. Our method aims to discover high-performance models that simultaneously optimize multiple objectives. We conduct comprehensive experiments on CIFAR-10, CIFAR-100, and ImageNet datasets to validate the effectiveness of our approach. With only a single GPU searching for 0.17 days, competitive architectures can be found by SMEMNAS, which achieves 78.91% accuracy with the MAdds of 570 M on the ImageNet. This work makes a significant advancement in the field of NAS.

On top of Segment Anything Model (SAM), SAM 2 further extends its capability from image to video inputs through a memory bank mechanism and obtains a remarkable performance compared with previous methods, making it a foundation model for video segmentation task. In this paper, we aim at making SAM 2 much more efficient so that it even runs on mobile devices while maintaining a comparable performance. Despite several works optimizing SAM for better efficiency, we find they are not sufficient for SAM 2 because they all focus on compressing the image encoder, while our benchmark shows that the newly introduced memory attention blocks are also the latency bottleneck. Given this observation, we propose EdgeTAM, which leverages a novel 2D Spatial Perceiver to reduce the computational cost. In particular, the proposed 2D Spatial Perceiver encodes the densely stored frame-level memories with a lightweight Transformer that contains a fixed set of learnable queries. Given that video segmentation is a dense prediction task, we find preserving the spatial structure of the memories is essential so that the queries are split into global-level and patch-level groups. We also propose a distillation pipeline that further improves the performance without inference overhead. As a result, EdgeTAM achieves 87.7, 70.0, 72.3, and 71.7 ${\mathcal{J}}{{\& }}{\mathcal{F}}$ on DAVIS 2017, MOSE, SA-V val, and SA-V test, while running at 16 FPS on iPhone 15 Pro Max. The code and models are available here.

Segment Anything Model 2 (SAM 2) has emerged as a powerful tool for video object segmentation and tracking anything. Key components of SAM 2 that drive the impressive video object segmentation performance include a large multistage image encoder for frame feature extraction and a memory mechanism that stores memory contexts from past frames to help current frame segmentation. The high computation complexity of multistage image encoder and memory module has limited its applications in real-world tasks, e.g., video object segmentation on mobile devices. To address this limitation, we propose EfficientTAMs, lightweight track anything models that produce high-quality results with low latency and model size. Our idea is based on revisiting the plain, nonhierarchical Vision Transformer (ViT) as an image encoder for video object segmentation, and introducing an efficient memory module, which reduces the complexity for both frame feature extraction and memory computation for current frame segmentation. We take vanilla lightweight ViTs and efficient memory module to build EfficientTAMs, and train the models on SA-1B and SA-V datasets for video object segmentation and track anything tasks. We evaluate on multiple video segmentation benchmarks including semi-supervised VOS and promptable video segmentation, and find that our proposed EfficientTAM with vanilla ViT perform comparably to SAM 2 model (HieraB+SAM 2) with ~2x speedup on A100 and ~2.4x parameter reduction. On segment anything image tasks, our EfficientTAMs also perform favorably over original SAM with ~20x speedup on A100 and ~20x parameter reduction. On mobile devices such as iPhone 15 Pro Max, our EfficientTAMs can run at ~10 FPS for performing video object segmentation with reasonable quality, highlighting the capability of small models for on-device video object segmentation applications.

Multimodal Large Language Models (MLLMs) have shown promising progress in understanding and analyzing video content. However, processing long videos remains a significant challenge constrained by LLM's context size. To address this limitation, we propose \textbf{LongVU}, a spatiotemporal adaptive compression mechanism that reduces the number of video tokens while preserving visual details of long videos. Our idea is based on leveraging cross-modal query and inter-frame dependencies to adaptively reduce temporal and spatial redundancy in videos. Specifically, we leverage DINOv2 features to remove redundant frames that exhibit high similarity. Then we utilize text-guided cross-modal query for selective frame feature reduction. Further, we perform spatial token reduction across frames based on their temporal dependencies. Our adaptive compression strategy effectively processes a large number of frames with little visual information loss within given context length. Our LongVU consistently surpass existing methods across a variety of video understanding benchmarks, especially on hour-long video understanding tasks such as VideoMME and MLVU. Given a light-weight LLM, our LongVU also scales effectively into a smaller size with state-of-the-art video understanding performance.

Segment Anything Model (SAM) has emerged as a powerful tool for numerous vision applications. A key component that drives the impressive performance for zero-shot trans-fer and high versatility is a super large Transformer model trained on the extensive high-quality SA -1 B dataset. While beneficial, the huge computation cost of SAM model has limited its applications to wider real-world applications. To address this limitation, we propose EfficientSAMs, light-weight SAM models that exhibits decent performance with largely reduced complexity. Our idea is based on leveraging masked image pretraining, SAMI, which learns to reconstruct features from SAM image encoder for effective visual representation learning. Further, we take SAMI-pretrained light-weight image encoders and mask decoder to build Effi-cientSAMs, and finetune the models on SA -1B for segment anything task. We perform evaluations on multiple vision tasks including image classification, object detection, in-stance segmentation, and semantic segmentation, and find that our proposed pretraining method, SAMI, consistently outperforms other masked image pretraining methods. On segment anything task such as zero-shot instance segmentation, our EfficientSAMs with SAMI-pretrained lightweight image encoders perform favorably with a significant gain (e.g., rv4 AP on COCOILVIS) over other fast SAM models. Our EfficientSAM code and models are available at here.

This paper investigates a phenomenon where query-based object detectors mispredict at the last decoding stage while predicting correctly at an intermediate stage. We review the training process and attribute the overlooked phenomenon to two limitations: lack of training emphasis and cascading errors from decoding sequence. We design and present Selective Query Recollection (SQR), a simple and effective training strategy for query-based object detectors. It cumulatively collects intermediate queries as decoding stages go deeper and selectively forwards the queries to the downstream stages aside from the sequential structure. Such-wise, SQR places training emphasis on later stages and allows later stages to work with intermediate queries from earlier stages directly. SQR can be easily plugged into various query-based object detectors and significantly enhances their performance while leaving the inference pipeline unchanged. As a result, we apply SQR on Adamixer, DAB-DETR, and Deformable-DETR across various settings (backbone, number of queries, schedule) and consistently brings 1.4 ~2.8 AP improvement. Code is available at https://github.com/Fangyi-Chen/SQR

Semantic segmentation is a crucial task in computer vision that involves segmenting images into semantically meaningful regions at the pixel level. However, existing approaches often rely on expensive human annotations as supervision for model training, limiting their scalability to large, unlabeled datasets. To address this challenge, we present ZeroSeg, a novel method that leverages the existing pretrained vision-language (VL) model (e.g. CLIP vision encoder [39]) to train open-vocabulary zero-shot semantic segmentation models. Although acquired extensive knowledge of visual concepts, it is non-trivial to exploit knowledge from these VL models to the task of semantic segmentation, as they are usually trained at an image level. ZeroSeg overcomes this by distilling the visual concepts learned by VL models into a set of segment tokens, each summarizing a localized region of the target image. We evaluate ZeroSeg on multiple popular segmentation benchmarks, including PASCAL VOC 2012, PASCAL Context, and COCO, in a zero-shot manner Our approach achieves state-of-the-art performance when compared to other zero-shot segmentation methods under the same training data, while also performing competitively compared to strongly supervised methods. Finally, we also demonstrated the effectiveness of ZeroSeg on open-vocabulary segmentation, through both human studies and qualitative visualizations. The code is publicly available at https://github.com/facebookresearch/ZeroSeg

Object detection has been expanded from a limited number of categories to open vocabulary. Moving forward, a complete intelligent vision system requires understanding more fine-grained object descriptions, object parts. In this paper, we propose a detector with the ability to predict both open-vocabulary objects and their part segmentation. This ability comes from two designs. First, we train the detector on the joint of part-level, object-level and image-level data to build the multi-granularity alignment between language and image. Second, we parse the novel object into its parts by its dense semantic correspondence with the base object. These two designs enable the detector to largely benefit from various data sources and foundation models. In open-vocabulary part segmentation experiments, our method outperforms the baseline by 3.3∼7.3 mAP in cross-dataset generalization on PartImageNet, and improves the baseline by 7.3 novel AP50in cross-category generalization on Pascal Part. Finally, we train a detector that generalizes to a wide range of part segmentation datasets while achieving better performance than dataset-specific training.

Object understanding in egocentric visual data is arguably a fundamental research topic in egocentric vision. However, existing object datasets are either non-egocentric or have limitations in object categories, visual content, and annotation granularities. In this work, we introduce EgoObjects, a large-scale egocentric dataset for fine-grained object understanding. Its Pilot version contains over 9K videos collected by 250 participants from 50+ countries using 4 wearable devices, and over 650K object annotations from 368 object categories. Unlike prior datasets containing only object category labels, EgoObjects also annotates each object with an instance-level identifier, and includes over 14K unique object instances. EgoObjects was designed to capture the same object under diverse background complexities, surrounding objects, distance, lighting and camera motion. In parallel to the data collection, we conducted data annotation by developing a multistage federated annotation process to accommodate the growing nature of the dataset. To bootstrap the research on EgoObjects, we present a suite of 4 benchmark tasks around the egocentric object understanding, including a novel instance level- and the classical category level object detection. Moreover, we also introduce 2 novel continual learning object detection tasks. The dataset and API are available at https://github.com/facebookresearch/EgoObjects.

Few-shot object detection is an imperative and long-lasting problem due to the inherent long-tail distribution of real-world data. Its performance is largely affected by the data scarcity of novel classes. But the semantic relation between the novel classes and the base classes is constant regardless of the data availability. In this work, we investigate utilizing this semantic relation together with the visual information and introduce explicit relation reasoning into the learning of novel object detection. Specifically, we represent each class concept by a semantic embedding learned from a large corpus of text. The detector is trained to project the image representations of objects into this embedding space. We also identify the problems of trivially using the raw embeddings with a heuristic knowledge graph and propose to augment the embeddings with a dynamic relation graph. As a result, our few-shot detector, termed SRR-FSD, is robust and stable to the variation of shots of novel objects. Experiments show that SRR-FSD can achieve competitive results at higher shots, and more importantly, a significantly better performance given both lower explicit and implicit shots. The benchmark protocol with implicit shots removed from the pretrained classification dataset can serve as a more realistic setting for future research.

This paper focuses on a novel and challenging detection scenario: A majority of true objects/instances is unlabeled in the datasets, so these missing-labeled areas will be regarded as the background during training. Previous art [1] on this problem has proposed to use soft sampling to re-weight the gradients of RoIs based on the overlaps with positive instances, while their method is mainly based on the two-stage detector (i.e. Faster RCNN) which is more robust and friendly for the missing label scenario. In this paper, we introduce a superior solution called Background Recalibration Loss (BRL) that can automatically re-calibrate the loss signals according to the pre-defined IoU threshold and input image. Our design is built on the one-stage detector which is faster and lighter. Inspired by the Focal Loss [2] formulation, we make several significant modifications to fit on the missing-annotation circumstance. We conduct extensive experiments on the curated PASCAL VOC [3] and MS COCO [4] datasets. The results demonstrate that our proposed method outperforms the baseline and other state-of-the-arts by a large margin.

We present simple and flexible drop-in modules in feature pyramids for general object detection, which can be easily generalized to other anchor-free detectors without introducing extra parameters, and only involves negligible computational cost on training and testing. The proposed detector, called NCMS, inserts a simple norm calibration (NC) operation between the feature pyramids and detection head to alleviate and balance the norm bias caused by feature pyramid network (FPN). Furthermore, the NCMS leverages an enhanced multi-feature selective strategy (MS) during training to assign the ground-truth to particular feature pyramid levels as supervisions, in order to obtain more discriminative representation for objects. By generalizing to the state-of-the-art FSAF module (Zhu et al., 2019), our NCMS improves it by 1.6% on COCO val set without bells and whistles. The resulting best model achieves 44.0% mAP with single-model and single-scale testing, which is a fairly competitive result.

Large-scale object detection datasets (e.g., MS-COCO) try to define the ground truth bounding boxes as clear as possible. However, we observe that ambiguities are still introduced when labeling the bounding boxes. In this paper, we propose a novel bounding box regression loss for learning bounding box transformation and localization variance together. Our loss greatly improves the localization accuracies of various architectures with nearly no additional computation. The learned localization variance allows us to merge neighboring bounding boxes during non-maximum suppression (NMS), which further improves the localization performance. On MS-COCO, we boost the Average Precision (AP) of VGG-16 Faster R-CNN from 23.6% to 29.1%. More importantly, for ResNet-50-FPN Mask R-CNN, our method improves the AP and AP90 by 1.8% and 6.2% respectively, which significantly outperforms previous state-of-the-art bounding box refinement methods. Our code and models are available at github.com/yihui-he/KL-Loss.

We motivate and present feature selective anchor-free (FSAF) module, a simple and effective building block for single-shot object detectors. It can be plugged into single-shot detectors with feature pyramid structure. The FSAF module addresses two limitations brought up by the conventional anchor-based detection: 1) heuristic-guided feature selection; 2) overlap-based anchor sampling. The general concept of the FSAF module is online feature selection applied to the training of multi-level anchor-free branches. Specifically, an anchor-free branch is attached to each level of the feature pyramid, allowing box encoding and decoding in the anchor-free manner at an arbitrary level. During training, we dynamically assign each instance to the most suitable feature level. At the time of inference, the FSAF module can work independently or jointly with anchor-based branches. We instantiate this concept with simple implementations of anchor-free branches and online feature selection strategy. Experimental results on the COCO detection track show that our FSAF module performs better than anchor-based counterparts while being faster. When working jointly with anchor-based branches, the FSAF module robustly improves the baseline RetinaNet by a large margin under various settings, while introducing nearly free inference overhead. And the resulting best model can achieve a state-of-the-art 44.6% mAP, outperforming all existing single-shot detectors on COCO.

Deep learning methods support vehicular technology in various aspects. How to efficiently and effectively optimize deep learning models remains a challenge. It is known that the learning rate is an important hyper-parameter to optimize models and the batch size is one of the keys to speed up training. This paper empirically studies the principles of scheduling batch size and learning rate during training, and proposes the Collaborative Control Strategy (CCS), which practically improves both classification accuracy and training speed. Instead of stepwise decreasing learning rate and keeping batch size unchanged, we asynchronously adjust batch size and learning rate based on time-division and restart policy. We study and analyze the proposed CCS on general image recognition benchmarks. Compared to traditional training strategy, without bells and whistles CCS decreases error rates by absolute 0.78% and 0.75% on CIFAR-10 and CIFAR-100 respectively, and reduces 40% of training time. For vehicular applications, we demonstrate its advantage on Stanford car-196 dataset with different architectures, showing consistent training speedup and accuracy improvement.

This paper introduces a novel anchor design principle to support anchor-based face detection for superior scaleinvariant performance, especially on tiny faces. To achieve this, we explicitly address the problem that anchor-based detectors drop performance drastically on faces with tiny sizes, e.g. less than 16 × 16 pixels. In this paper, we investigate why this is the case. We discover that current anchor design cannot guarantee high overlaps between tiny faces and anchor boxes, which increases the difficulty of training. The new Expected Max Overlapping (EMO) score is proposed which can theoretically explain the low overlapping issue and inspire several effective strategies of new anchor design leading to higher face overlaps, including anchor stride reduction with new network architectures, extra shifted anchors, and stochastic face shifting. Comprehensive experiments show that our proposed method significantly outperforms the baseline anchor-based detector, while consistently achieving state-of-the-art results on challenging face detection datasets with competitive runtime speed.

Although face detection has been intensely studied for decades, it is still a challenging topic due to numerous conditions, e.g. heavy occlusions, low resolutions, extreme poses, non-face patterns that look like human faces, etc. This paper proposes a novel region-based ConvNet to address these issues. Our approach enhances the interior deep facial features and explicitly incorporates the exterior deep features. The enhanced interior features provide fine details for small faces. The exterior features capture the local information surrounding the face, supporting the detection under challenging conditions. Experiments show that our proposed components improve the baseline method significantly. Additionally, our approach consistently achieves competitive performance in four challenging databases, i.e. Wider Face, AFW, PASCAL Faces, and FDDB. We also introduce a new challenging non-face dataset 1 of 6,000 images to benchmark false positive rates for future research.

The problems of hand detection have been widely addressed in many areas, e.g. human computer interaction environment, driver behaviors monitoring, etc. However, the detection accuracy in recent hand detection systems are still far away from the demands in practice due to a number of challenges, e.g. hand variations, highly occlusions, low-resolution and strong lighting conditions. This paper presents the Multiple Scale Faster Region-based Convolutional Neural Network (MS-FRCNN) to handle the problems of hand detection in given digital images collected under challenging conditions. Our proposed method introduces a multiple scale deep feature extraction approach in order to handle the challenging factors to provide a robust hand detection algorithm. The method is evaluated on the challenging hand database, i.e. the Vision for Intelligent Vehicles and Applications (VIVA) Challenge, and compared against various recent hand detection methods. Our proposed method achieves the state-of-the-art results with 20% of the detection accuracy higher than the second best one in the VIVA challenge.

Big data has been becoming ubiquitous and applied in numerous fields recently. The challenges to solve a large-scale machine learning problem in big data scenario generally lie in three aspects. Firstly, a proposed machine learning algorithm has to be appropriated for the distributed optimization problem. Secondly, it needs a platform for the distributed implementation. Finally, the communication delays different machines may cause problems in convergence even though the non-distributed algorithm shows a good convergence rate. In order to solve these challenges, we propose a new machine learning approach named Distributed Class-dependent Feature Analysis (DCFA), to combine the advantages of sparse representation in an over-complete dictionary. The classifier is based on the estimation of class-specific optimal filters, by solving an l1-norm optimization problem. We demonstrate how this problem is solved using the Alternating Direction Method of Multipliers and also explore relevant convergency details. More importantly, our proposed framework can be efficiently implemented on a robust distributed framework. Thus, it improves both accuracy and computational time in large-scale databases. Our method achieves very high classification accuracies in face recognition in the presence of occlusions on AR database. It also outperforms the state of the art methods in object recognition on two challenging large-scale object databases, i.e. Caltech101 and Caltech256. It hence shows its applicability to general computer vision and pattern recognition problems. In addition, computational time experiments show our distributed method achieves high speedup of 7.85x on Caltech256 databases with just 10 machine nodes compared to the non-distributed version and can gain even more with more computing resources.

Multifocus image fusion is an important research topic in the computer vision and image processing field. The optical lenses that are commonly used by imaging devices, such as auto-focus cameras, have a limiting focus range. Thus, only objects within the range of distances from the devices can be captured and recorded sharply while out-of-range objects become blur. In this paper, we present a novel image fusion scheme for combining two or multiple images with different focus points to generate an all-in-focus image. We formulate the problem of fusing multifocus images as choosing most significant features from a sparse matrix produced by a newly developed robust principal component analysis (RPCA) decomposition method to form a composite feature space. Thus, the salient features presented in sharp regions can be captured and integrated into a single representation. The sparse matrix is first divided into small blocks, and standard deviation is then calculated on each block as a selection criterion. To reduce blocking artifacts, a sliding window technique is utilized to smooth the transitions between blocks. The proposed fusion scheme has been demonstrated to successfully improve fusion quality in terms of visual and quantitative evaluations. The method is also able to effectively handle both grayscale and color images.