Today’s most advanced vision-language models (VLMs) remain proprietary. The strongest open-weight models rely heavily on synthetic data from proprietary VLMs to achieve good performance, effectively distilling these closed VLMs into open ones. As a result, the community has been missing foundational knowledge about how to build performant VLMs from scratch. We present Molmo, a new family of VLMs that are state-of-the-art in their class of openness. Our key contribution is a collection of new datasets called PixMo, including a dataset of highly detailed image captions for pre-training, a free-form image Q&A dataset for fine-tuning, and an innovative 2D pointing dataset, all collected without the use of external VLMs. The success of our approach relies on careful modeling choices, a well- tuned training pipeline, and, most critically, the quality of our newly collected datasets. Our best-in-class 72B model not only outperforms others in the class of open weight and data models, but also outperforms larger proprietary models including Claude 3.5 Sonnet, and Gemini 1.5 Pro and Flash, second only to GPT-4o based on both academic benchmarks and on a large human evaluation. Our model weights, new datasets, and source code are available at https://molmo.allenai.org/blog.

We introduce a highly multimodal transformer to represent many remote sensing modalities - multispectral optical, synthetic aperture radar, elevation, weather, pseudo-labels, and more - across space and time. These inputs are useful for diverse remote sensing tasks, such as crop mapping and flood detection. However, learning shared representations of remote sensing data is challenging, given the diversity of relevant data modalities, and because objects of interest vary massively in scale, from small boats (1-2 pixels and fast) to glaciers (thousands of pixels and slow). We present a novel self-supervised learning algorithm that extracts multi-scale features across a flexible set of input modalities through masked modeling. Our dual global and local contrastive losses differ in their targets (deep representations vs. shallow input projections) and masking strategies (structured vs. not). Our Galileo is a single generalist model that outperforms SoTA specialist models for satellite images and pixel time series across eleven benchmarks and multiple tasks.

In the face of pressing environmental issues in the 21st century, monitoring surface changes on Earth is more important than ever. Large-scale remote sensing, such as satellite imagery, is an important tool for this task. However, using supervised methods to detect changes is difficult because of the lack of satellite data annotated with change labels, especially for rare categories of change. Annotation proves challenging due to the sparse occurrence of changes in satellite images. Even within a vast collection of images, only a small fraction may exhibit persistent changes of interest. To address this challenge, we introduce OPTIMUS, a self-supervised learning method based on an intuitive principle: if a model can recover information about the relative order of images in the time series, then that implies that there are long-lasting changes in the images. OPTIMUS demonstrates this principle by using change point detection methods on model outputs in a time series. We demonstrate that OPTIMUS can directly detect interesting changes in satellite images, achieving an improvement in AUROC score from 56.3% to 87.6% at distinguishing changed time series from unchanged ones compared to baselines. Our code and dataset are available at https://huggingface.co/datasets/optimus-change/optimus-dataset/.

Abstract Data scientists often need to write programs to process predictions of machine learning models, such as object detections and trajectories in video data. However, writing such queries can be challenging due to the fuzzy nature of real-world data; in particular, they often include real-valued parameters that must be tuned by hand. We propose a novel framework calledQuivrthat synthesizes trajectory queries matching a given set of examples. To efficiently synthesize parameters, we introduce a novel technique for pruning the parameter space and a novel quantitative semantics that makes this more efficient. We evaluateQuivron a benchmark of 17 tasks, including several from prior work, and show both that it can synthesize accurate queries for each task and that our optimizations substantially reduce synthesis time.

Remote sensing images are useful for a wide variety of planet monitoring applications, from tracking deforestation to tackling illegal fishing. The Earth is extremely diverse—the amount of potential tasks in remote sensing images is massive, and the sizes of features range from several kilometers to just tens of centimeters. However, creating generalizable computer vision methods is a challenge in part due to the lack of a large-scale dataset that captures these diverse features for many tasks. In this paper, we present SatlasPretrain, a remote sensing dataset that is large in both breadth and scale, combining Sentinel-2 and NAIP images with 302M labels under 137 categories and seven label types. We evaluate eight baselines and a proposed method on SatlasPretrain, and find that there is substantial room for improvement in addressing research challenges specific to remote sensing, including processing image time series that consist of images from very different types of sensors, and taking advantage of long-range spatial context. Moreover, we find that pre-training on SatlasPretrain substantially improves performance on downstream tasks, increasing average accuracy by 18% over ImageNet and 6% over the next best baseline. The dataset, pre-trained model weights, and code are available at https://satlas-pretrain.allen.ai/.

Capturing and processing video is increasingly common as cameras become cheaper to deploy. At the same time, rich video-understanding methods have progressed greatly in the last decade. As a result, many organizations now have massive repositories of video data, with applications in mapping, navigation, autonomous driving, and other areas. Because state-of-the-art object-detection methods are slow and expensive, our ability to process even simple ad-hoc object search queries (“find 100 traffic lights in dashcam video”) over this accumulated data lags far behind our ability to collect the data. Processing video at reduced sampling rates is a reasonable default strategy for these types of queries; however, the ideal sampling rate is both data and query dependent. We introduce ExSample, a low cost framework for object search over un-indexed video that quickly processes search queries by adapting the amount and location of sampled frames to the particular data and query being processed. ExSample prioritizes the processing of frames in a video repository so that processing is focused in portions of video that most likely contain objects of interest. It approaches searching in a similar way to a multi-arm bandit problem where each arm corresponds to a portion of a video. On large, real-world datasets, ExSample reduces processing time by 1.9x on average and up to 6x over an efficient random sampling baseline. Moreover, we show ExSample finds many results long before sophisticated, state-of-the-art baselines based on proxy scores can begin producing their first results.

Capturing and processing video is increasingly com-mon as cameras become cheaper to deploy. At the same time, rich video-understanding methods have progressed greatly in the last decade. As a result, many organizations now have massive repositories of video data, with applications in mapping, navigation, autonomous driving, and other areas. Because state-of-the-art object-detection methods are slow and expensive, our ability to process even simple ad-hoc object search queries (“find 100 traffic lights in dashcam video”) over this accumulated data lags far behind our ability to collect the data. Processing video at reduced sampling rates is a reasonable default strategy for these types of queries; however, the ideal sampling rate is both data and query dependent. We introduce ExSample, a low cost framework for object search over un-indexed video that quickly processes search queries by adapting the amount and location of sampled frames to the particular data and query being processed. ExSample prioritizes the processing of frames in a video repository so that processing is focused in portions of video that most likely contain objects of interest. It approaches searching in a similar way to a multi-arm bandit problem where each arm corresponds to a portion of a video. On large, real-world datasets, ExSample reduces processing time by 1.9x on average and up to 6x over an efficient random sampling baseline. Moreover, we show ExSample finds many results long before sophisticated, state-of-the-art baselines based on proxy scores can begin producing their first results.

Performing analytics tasks over large-scale video datasets is increasingly common in a wide range of applications, from traffic planning to sports analytics. These tasks generally involve object detection and tracking operations that require pre-processing the video through expensive machine learning models. To address this cost, several video query optimizers have recently been proposed. Broadly, these methods trade large reductions in pre-processing cost for increases in query execution cost: during query execution, they apply query-specific machine learning operations over portions of the video dataset. Although video query optimizers reduce the overall cost of executing a single query over large video datasets compared to naive object tracking methods, executing several queries over the same video remains cost-prohibitive; moreover, the high per-query latency makes these systems unsuitable for exploratory analytics where fast response times are crucial.

Accurately maintaining digital street maps is labor-intensive. To address this challenge, much work has studied automatically processing geospatial data sources such as GPS trajectories and satellite images to reduce the cost of maintaining digital maps. An end-to-end map update system would first process geospatial data sources to extract insights, and second leverage those insights to update and improve the map. However, prior work largely focuses on the first step of this pipeline: these map extraction methods infer road networks from scratch given geospatial data sources (in effect creating entirely new maps), but do not address the second step of leveraging this extracted information to update the existing map data. In this paper, we first explain why current map extraction techniques yield low accuracy when extended to update existing maps. We then propose a novel method that leverages the progression of satellite imagery over time to substantially improve accuracy. Our approach first compares satellite images captured at different times to identify portions of the physical road network that have visibly changed, and then updates the existing map accordingly. We show that our change-based approach reduces error rates four-fold.

The increasing availability of satellite and aerial imagery has sparked substantial interest in automatically updating street maps by processing aerial images. Until now, the community has largely focused on road extraction, where road networks are inferred from scratch from an aerial image. However, given that relatively high-quality maps exist in most parts of the world, in practice, inference approaches must be applied to update existing maps rather than infer new ones. With recent road extraction methods showing high accuracy, we argue that it is time to transition to the more practical map update task, where an existing map is updated by adding, removing, and shifting roads, without introducing errors in parts of the existing map that remain up-to-date. In this paper, we develop a new dataset called MUNO21 for the map update task, and show that it poses several new and interesting research challenges. We evaluate several state-of-the-art road extraction methods on MUNO21, and find that substantial further improvements in accuracy will be needed to realize automatic map update.

In this paper, we propose a self-supervised learning procedure for training a robust multi-object tracking (MOT) model given only unlabeled video. While several self-supervisory learning signals have been proposed in prior work on single-object tracking, such as color propagation and cycle-consistency, these signals are not effective for training RNN models, which are needed to achieve accurate MOT: they yield degenerate models that, for instance, always match new detections to tracks with the closest initial detections. We propose a novel self-supervisory signal that we call cross-input consistency: we construct two distinct inputs for the same sequence of video, by hiding different information about the sequence in each input. We then compute tracks in that sequence by applying an RNN model independently on each input, and train the model to produce consistent tracks across the two inputs. We evaluate our unsupervised method on MOT17 and KITTI --- remarkably, we find that, despite training only on unlabeled video, our unsupervised approach outperforms four supervised methods published in the last 1--2 years, including Tracktor++, FAMNet, GSM, and mmMOT.

Inferring road attributes such as lane count and road type from satellite imagery is challenging. Often, due to the occlusion in satellite imagery and the spatial correlation of road attributes, a road attribute at one position on a road may only be apparent when considering far-away segments of the road. Thus, to robustly infer road attributes, the model must integrate scattered information and capture the spatial correlation of features along roads. Existing solutions that rely on image classifiers fail to capture this correlation, resulting in poor accuracy. We find this failure is caused by a fundamental limitation – the limited effective receptive field of image classifiers.To overcome this limitation, we propose RoadTagger, an end-to-end architecture which combines both Convolutional Neural Networks (CNNs) and Graph Neural Networks (GNNs) to infer road attributes. Using a GNN allows information to propagate on the road network graph and eliminates the receptive field limitation of image classifiers. We evaluate RoadTagger on both a large real-world dataset covering 688 km2 area in 20 U.S. cities and a synthesized dataset. In the evaluation, RoadTagger improves inference accuracy over the CNN image classifier based approaches. In addition, RoadTagger is robust to disruptions in the satellite imagery and is able to learn complicated inductive rules for aggregating scattered information along the road network.

The rapid development of small aerial drones has enabled numerous drone-based applications, e.g., geographic mapping, air pollution sensing, and search and rescue. To assist the development of these applications, we propose BeeCluster, a drone orchestration system that manages a fleet of drones. BeeCluster provides a virtual drone abstraction that enables developers to express a sequence of geographical sensing tasks, and determines how to map these tasks to the fleet efficiently. BeeCluster's core contribution is predictive optimization, in which an inferred model of the future tasks of the application is used to generate an optimized flight and sensing schedule for the drones that aims to minimize the total expected execution time.

Video databases that enable queries with object-track predicates are useful in many applications. Such queries include selecting objects that move from one region of the camera frame to another (e.g., finding cars that turn right through a junction) and selecting objects with certain speeds (e.g., finding animals that stop to drink water from a lake). Processing such predicates efficiently is challenging because they involve the movement of an object over several video frames. We propose a novel query-driven tracking approach that integrates query processing with object tracking to efficiently process object track queries and address the computational complexity of object detection methods. By processing video at low framerates when possible, but increasing the framerate when needed to ensure high-accuracy on a query, our approach substantially speeds up query execution. We have implemented query-driven tracking in MIRIS, a video query processor, and compare MIRIS against four baselines on a diverse dataset consisting of five sources of video and nine distinct queries. We find that, at the same accuracy, MIRIS accelerates video query processing by 9x on average over the IOU tracker, an overlap-based tracking-by-detection method used in existing video database systems.

We demonstrate Vaas, a video analytics system for large-scale datasets. Vaas provides an interactive interface to rapidly develop and experiment with different workflows for solving a video analytics task. Users express these workflows as Vaas queries, which specify data flow graphs where nodes may be implemented by machine learning models, custom code, or basic built-in operations (e.g., cropping, selecting detections of by class, filtering tracks by bounding boxes). For example, the problem of detecting lane change events in dashboard camera video could be solved directly as an activity recognition task, by training a model to classify whether a segment of video contains a lane change, or decomposed into a set of simpler tasks, such as detecting lane markers and then identifying shifts in the detected lanes. Our system interface incorporates a query composition tool, where users can rapidly compose operations to implement a workflow, and an exploration tool, where users can experiment with a query by applying it over samples from the dataset to fix bugs and tune parameters. Vaas incorporates recent work in approximate video query processing to support the fast, interactive execution of queries, and accelerates the annotation process of hand-labeling examples to train models by allowing users to annotate over the outputs of previously expressed queries rather than the entire video dataset.

Mapping road networks is currently both expensive and labor-intensive. High-resolution aerial imagery provides a promising avenue to automatically infer a road network. Prior work uses convolutional neural networks (CNNs) to detect which pixels belong to a road (segmentation), and then uses complex post-processing heuristics to infer graph connectivity. We show that these segmentation methods have high error rates because noisy CNN outputs are difficult to correct. We propose RoadTracer, a new method to automatically construct accurate road network maps from aerial images. RoadTracer uses an iterative search process guided by a CNN-based decision function to derive the road network graph directly from the output of the CNN. We compare our approach with a segmentation method on fifteen cities, and find that at a 5% error rate, RoadTracer correctly captures 45% more junctions across these cities.

Mapping road networks today is labor-intensive. As a result, road maps have poor coverage outside urban centers in many countries. Systems to automatically infer road network graphs from aerial imagery and GPS trajectories have been proposed to improve coverage of road maps. However, because of high error rates, these systems have not been adopted by mapping communities. We propose machine-assisted map editing, where automatic map inference is integrated into existing, human-centric map editing workflows. To realize this, we build Machine-Assisted iD (MAiD), where we extend the web-based OpenStreetMap editor, iD, with machine-assistance functionality. We complement MAiD with a novel approach for inferring road topology from aerial imagery that combines the speed of prior segmentation approaches with the accuracy of prior iterative graph construction methods. We design MAiD to tackle the addition of major, arterial roads in regions where existing maps have poor coverage, and the incremental improvement of coverage in regions where major roads are already mapped. We conduct two user studies and find that, when participants are given a fixed time to map roads, they are able to add as much as 3.5x more roads with MAiD.

Current approaches to construct road network maps from GPS trajectories suffer from low precision, especially in dense urban areas and in regions with complex topologies such as overpasses and underpasses, parallel roads, and stacked roads. This paper proposes a two-stage method to improve precision without sacrificing recall (coverage). The first stage, RoadRunner, is a method that can generate high-precision maps even in challenging scenarios by incrementally following the flow of trajectories, using the connectivity between observations in each trajectory to decide whether overlapping trajectories are traversing the same road or distinct parallel roads, and to correctly infer road segment connectivity. By itself, RoadRunner is not designed to achieve high recall, but we show how to combine it with a wide range of prior schemes, some that use GPS trajectories and some that use aerial imagery, to achieve recall similar to prior schemes but at substantially higher precision. We evaluated RoadRunner in four U.S. cities using 60,000 GPS trajectories, and found that precision improves by 5.2 points (a 33.6% error rate reduction) and 24.3 points (a 60.7% error rate reduction) over two existing schemes, with a slight increase in recall.

Urban traffic gridlock is a familiar scene. At the same time, the mean occupancy rate of personal vehicle trips in the United States is only 1.6 persons per vehicle mile. Ridesharing has the potential to solve many environmental, congestion, pollution, and energy problems. In this paper, we introduce the problem of large scale real-time ridesharing with service guarantee on road networks. Trip requests are dynamically matched to vehicles while trip waiting and service time constraints are satisfied. We first propose two scheduling algorithms: a branch-and-bound algorithm and an integer programing algorithm. However, these algorithms do not adapt well to the dynamic nature of the ridesharing problem. Thus, we propose kinetic tree algorithms which are better suited to efficient scheduling of dynamic requests and adjust routes on-the-fly. We perform experiments on a large Shanghai taxi dataset. Results show that the kinetic tree algorithms outperform other algorithms significantly.

This demo presents Noah: a dynamic ridesharing system. Noah supports large scale real-time ridesharing with service guarantee on road networks. Taxis and trip requests are dynamically matched. Different from traditional systems, a taxi can have more than one customer on board given that all waiting time and service time constraints of trips are satisfied. Noah's real-time response relies on three main components: (1) a fast shortest path algorithm with caching on road networks; (2) fast dynamic matching algorithms to schedule ridesharing on the fly; (3) a spatial indexing method for fast retrieving moving taxis. Users will be able to submit requests from a smartphone, choose specific parameters such as number of taxis in the system, service constraints, and matching algorithms, to explore the internal functionalities and implementations of Noah. The system analyzer will show the system performance including average waiting time, average detour percentage, average response time, and average level of sharing. Taxis, routes, and requests will be animated and visualized through Google Maps API. The demo is based on trips of 17,000 Shanghai taxis for one day (May 29, 2009); the dataset contains 432,327 trips. Each trip includes the starting and destination coordinates and the start time. An iPhone application is implemented to allow users to submit a trip request to the Noah system during the demonstration.

Service-oriented architecture provides a framework for achieving rapid system composition and deployment. To satisfy different system QoS requirements, it is possible to select an appropriate set of concrete services and compose them to achieve the QoS goals. In addition, some of the services may be reconfigurable and provide various QoS tradeoffs. To make use of these reconfigurable services, the composition process should consider not only service selection, but also configuration parameter settings. However, existing QoS-driven service composition research does not consider reconfigurable services. Moreover, the decision space may be enormous when reconfigurable services are considered. In this paper, we deal with the issues of reconfigurable service modeling and efficient service composition decision making. We introduce a novel compositional decision making process, CDP, which explores optimal solutions of individual component services and uses the knowledge to derive optimal QoS-driven composition solutions. Experimental studies show that the CDP approach can significantly reduce the search space and achieve great performance gains. We also develop a case study system to validate the proposed approach and the results confirm the feasibility and effectiveness of reconfigurable services.

We propose a flexible mini-shuttle like transportation system called flexi, with routes formed by analyzing passenger trip data from a large set of taxi trajectories. The usage of public transportation is declining as often it no longer matches with individual needs. Thus, the flexi system provides a transportation mode in between buses and taxis so that inconvenience in switching to the system can be minimized overall. To generate flexi routes, we propose a two-phase approach. In the first phase, a fast diameter-constrained agglomerative clustering algorithm is developed and applied to the set of trips derived from the GPS data. This phase identifies a set of heavily traveled spatio-temporal trip clusters called hot lines. In the second phase, a directed acyclic graph is constructed from the hot lines. Then, an optimal single flexi route discovery algorithm on graph searching is proposed. Multiple routes are discovered by iteratively applying the single routing algorithm. Extensive experiments using a large set of real taxi trajectory data show that the flexi system can save a large percentage of trip mileage.