Reconstructing human-object interactions (HOI) from single images is fundamental in computer vision. Existing methods are primarily trained and tested on indoor scenes due to the lack of 3D data, particularly constrained by the object variety, making it challenging to generalize to real-world scenes with a wide range of objects. The limitations of previous 3D HOI datasets were primarily due to the difficulty in acquiring 3D object assets. However, with the development of 3D reconstruction from single images, recently it has become possible to reconstruct various objects from 2D HOI images. We therefore propose a pipeline for annotating fine-grained 3D humans, objects, and their interactions from single images. We annotated 2.5k+ 3D HOI assets from existing 2D HOI datasets and built the first open-vocabulary in-the-wild 3D HOI dataset Open3DHOI, to serve as a future test set. Moreover, we design a novel Gaussian-HOI optimizer, which efficiently reconstructs the spatial interactions between humans and objects while learning the contact regions. Besides the 3D HOI reconstruction, we also propose several new tasks for 3D HOI understanding to pave the way for future work. Data and code will be publicly available at https://wenboran2002.github.io/3dhoi/.

Online display advertising platforms service numerous advertisers by providing real-time bidding (RTB) for the scale of billions of ad requests every day. The bidding strategy handles ad requests cross multiple channels to maximize the number of clicks under the set financial constraints, i.e., total budget and cost-per-click (CPC), etc. Different from existing works mainly focusing on single channel bidding, we explicitly consider cross-channel constrained bidding with budget allocation. Specifically, we propose a hierarchical offline deep reinforcement learning (DRL) framework called “HiBid”, consisted of a high-level planner equipped with auxiliary loss for non-competitive budget allocation, and a data augmentation enhanced low-level executor for adaptive bidding strategy in response to allocated budgets. Additionally, a CPC-guided action selection mechanism is introduced to satisfy the cross-channel CPC constraint. Through extensive experiments on both the large-scale log data and online A/B testing, we confirm that HiBid outperforms six baselines in terms of the number of clicks, CPC satisfactory ratio, and return-on-investment (ROI). We also deploy HiBid on Meituan advertising platform to already service tens of thousands of advertisers every day.

Recommender systems aim to recommend the most suitable items to users from a large number of candidates. Their computation cost grows as the number of user requests and the complexity of services (or models) increases. Under the limitation of computation resources (CRs), how to make a trade-off between computation cost and business revenue becomes an essential question. The existing studies focus on dynamically allocating CRs in queue truncation scenarios (i.e., allocating the size of candidates), and formulate the CR allocation problem as an optimization problem with constraints. Some of them focus on single-phase CR allocation, and others focus on multi-phase CR allocation but introduce some assumptions about queue truncation scenarios. However, these assumptions do not hold in other scenarios, such as retrieval channel selection and prediction model selection. Moreover, existing studies ignore the state transition process of requests between different phases, limiting the effectiveness of their approaches.

Docker is the de-facto container technology for software system deployment and delivery. A Dockerfile specifies how to containerize a system into a Docker image. However, creating a Dockerfile is not trivial since resolving the dependencies (e.g., third-party libraries) of diverse software requires comprehensive domain knowledge. In this paper, we propose DockerGen to containerize software packages automatically. DockerGen constructs a knowledge graph containing rich knowledge of building Docker images by analyzing nearly 220 thousand Dockerfiles. DockerGen exploits the knowledge graph to containerize the target software by creating a Dockerfile specifying the base image, dependencies, and the operation workflow. We evaluate DockerGen on 100 software packages of various categories. DockerGen achieves a 73% build success rate and a 59% configuration success rate. The experimental result indicates it is viable to automate software containerization based on a domain knowledge graph.

Docker images, having the idea of "build once, run anywhere", are widely used as the reusable delivery artifacts. Currently, there are a huge number of online Docker repositories that provide images as the off-the-shelf blocks to construct large and complicated systems. Tags would improve the reusability as they provide concise semantics. However, tags are not well supported for Docker images, and manual tagging is still exhausting. We propose STAR, a Specialized Tagging Approach for Docker Repositories, to address the problem of automatically multi-labeling the large number of repositories. STAR takes Dockerfiles of the repositories as the primary input, which because a Dockerfile contains all the instructions for building a Docker image. STAR is based on two prediction models. By taking a Dockerfile as the specific text description, we model a repository with its labeled tags and Dockerfile terms, and use Labeled Latent Dirichlet Allocation algorithm to recommend tags. By regarding a Dockerfile as the configuration code, we construct a feature model based on Dockerfile key instructions and use a similarity-based ranking algorithm to recommend tags. Given an untagged repository, STAR outputs two probability scores for each tag with the two models and takes a weighted sum of them as the final score. Finally, STAR ranks all the tags according to their scores and recommends the top K ones. We evaluate STAR on over 100,000 repositories of Docker Hub. The experimental results show that STAR outperforms the state-of-the-art approaches in terms of Recall@5 and Recall@10.

Docker repositories usually contain Docker images and Dockerfiles, where Docker images are a kind of off-the-shelf artifact and Dockerfiles specify how to automatically build Docker images following the notion of Infrastructure-as-Code. Given a huge number of Docker repositories, tag recommendation is essential to ensure that relevant ones can be easily retrieved, because tagging is practical in describing, bookmarking, navigating and searching software objects. However, in Docker Hub, tags are not well supported to semantically describing the repositories, and manually tagging is still an exhausting and time-consuming task.