GTA: A Benchmark for General Tool Agents

Significant focus has been placed on integrating large language models (LLMs) with various tools in developing general-purpose agents. This poses a challenge to LLMs' tool-use capabilities. However, there are evident gaps between existing tool-use evaluations and real-world scenarios. Current evaluations often use AI-generated queries, single-step tasks, dummy tools, and text-only interactions, failing to reveal the agents' real-world problem-solving abilities effectively. To address this, we propose GTA, a benchmark for General Tool Agents, featuring three main aspects: (i) Real user queries: human-written queries with simple real-world objectives but implicit tool-use, requiring the LLM to reason the suitable tools and plan the solution steps. (ii) Real deployed tools: an evaluation platform equipped with tools across perception, operation, logic, and creativity categories to evaluate the agents' actual task execution performance. (iii) Real multimodal inputs: authentic image files, such as spatial scenes, web page screenshots, tables, code snippets, and printed/handwritten materials, used as the query contexts to align with real-world scenarios closely. We design 229 real-world tasks and executable tool chains to evaluate mainstream LLMs. Our findings show that real-world user queries are challenging for existing LLMs, with GPT-4 completing less than 50% of the tasks and most LLMs achieving below 25%. This evaluation reveals the bottlenecks in the tool-use capabilities of current LLMs in real-world scenarios, which provides future direction for advancing general-purpose tool agents. The code and dataset are available at https://github.com/open-compass/GTA.

ADTR: Anomaly Detection Transformer with Feature Reconstruction

Anomaly detection with only prior knowledge from normal samples attracts more attention because of the lack of anomaly samples. Existing CNN-based pixel reconstruction approaches suffer from two concerns. First, the reconstruction source and target are raw pixel values that contain indistinguishable semantic information. Second, CNN tends to reconstruct both normal samples and anomalies well, making them still hard to distinguish. In this paper, we propose Anomaly Detection TRansformer (ADTR) to apply a transformer to reconstruct pre-trained features. The pre-trained features contain distinguishable semantic information. Also, the adoption of transformer limits to reconstruct anomalies well such that anomalies could be detected easily once the reconstruction fails. Moreover, we propose novel loss functions to make our approach compatible with the normal-sample-only case and the anomaly-available case with both image-level and pixel-level labeled anomalies. The performance could be further improved by adding simple synthetic or external irrelevant anomalies. Extensive experiments are conducted on anomaly detection datasets including MVTec-AD and CIFAR-10. Our method achieves superior performance compared with all baselines.

A Unified Model for Multi-class Anomaly Detection

Despite the rapid advance of unsupervised anomaly detection, existing methods require to train separate models for different objects. In this work, we present UniAD that accomplishes anomaly detection for multiple classes with a unified framework. Under such a challenging setting, popular reconstruction networks may fall into an "identical shortcut", where both normal and anomalous samples can be well recovered, and hence fail to spot outliers. To tackle this obstacle, we make three improvements. First, we revisit the formulations of fully-connected layer, convolutional layer, as well as attention layer, and confirm the important role of query embedding (i.e., within attention layer) in preventing the network from learning the shortcut. We therefore come up with a layer-wise query decoder to help model the multi-class distribution. Second, we employ a neighbor masked attention module to further avoid the information leak from the input feature to the reconstructed output feature. Third, we propose a feature jittering strategy that urges the model to recover the correct message even with noisy inputs. We evaluate our algorithm on MVTec-AD and CIFAR-10 datasets, where we surpass the state-of-the-art alternatives by a sufficiently large margin. For example, when learning a unified model for 15 categories in MVTec-AD, we surpass the second competitor on the tasks of both anomaly detection (from 88.1% to 96.5%) and anomaly localization (from 89.5% to 96.8%). Code will be made publicly available.

The Z-axis, X-axis, Weight and Disambiguation Methods for Constructing Local Reference Frame in 3D Registration: An Evaluation

The local reference frame (LRF), as an independent coordinate system generated on a local 3D surface, is widely used in 3D local feature descriptor construction and 3D transformation estimation which are two key steps in the local method-based surface matching. There are numerous LRF methods have been proposed in literatures. In these methods, the x- and z-axis are commonly generated by different methods or strategies, and some x-axis methods are implemented on the basis of a z-axis being given. In addition, the weight and disambiguation methods are commonly used in these LRF methods. In existing evaluations of LRF, each LRF method is evaluated with a complete form. However, the merits and demerits of the z-axis, x-axis, weight and disambiguation methods in LRF construction are unclear. In this paper, we comprehensively analyze the z-axis, x-axis, weight and disambiguation methods in existing LRFs, and obtain six z-axis and eight x-axis, five weight and two disambiguation methods. The performance of these methods are comprehensively evaluated on six standard datasets with different application scenarios and nuisances. Considering the evaluation outcomes, the merits and demerits of different weight, disambiguation, z- and x-axis methods are analyzed and summarized. The experimental result also shows that some new designed LRF axes present superior performance compared with the state-of-the-art ones.

Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels

The crux of semi-supervised semantic segmentation is to assign adequate pseudo-labels to the pixels of unlabeled images. A common practice is to select the highly confident predictions as the pseudo ground-truth, but it leads to a problem that most pixels may be left unused due to their unreliability. We argue that every pixel matters to the model training, even its prediction is ambiguous. Intuitively, an unreliable prediction may get confused among the top classes (i.e., those with the highest probabilities), however, it should be confident about the pixel not belonging to the remaining classes. Hence, such a pixel can be convincingly treated as a negative sample to those most unlikely categories. Based on this insight, we develop an effective pipeline to make sufficient use of unlabeled data. Concretely, we separate reliable and unreliable pixels via the entropy of predictions, push each unreliable pixel to a category-wise queue that consists of negative samples, and manage to train the model with all candidate pixels. Considering the training evolution, where the prediction becomes more and more accurate, we adaptively adjust the threshold for the reliable-unreliable partition. Experimental results on various benchmarks and training settings demonstrate the superiority of our approach over the state-of-the-art alternatives.

Iterative Correlation-based Feature Refinement for Few-shot Counting

Few-shot counting aims to count objects of any class in an image given only a few exemplars of the same class. Existing correlation-based few-shot counting approaches suffer from the coarseness and low semantic level of the correlation. To solve these problems, we propose an iterative framework to progressively refine the exemplar-related features based on the correlation between the image and exemplars. Then the density map is predicted from the final refined feature map. The iterative framework includes a Correlation Distillation module and a Feature Refinement module. During the iterations, the exemplar-related features are gradually refined, while the exemplar-unrelated features are suppressed, benefiting few-shot counting where the exemplar-related features are more important. Our approach surpasses all baselines significantly on few-shot counting benchmark FSC-147. Surprisingly, though designed for general class-agnostic counting, our approach still achieves state-of-the-art performance on car counting benchmarks CARPK and PUCPR+, and crowd counting benchmarks UCSD and Mall. We also achieve competitive performance on crowd counting benchmark ShanghaiTech. The code will be released soon.

PF-Net: Point Fractal Network for 3D Point Cloud Completion

In this paper, we propose a Point Fractal Network (PF-Net), a novel learning-based approach for precise and high-fidelity point cloud completion. Unlike existing point cloud completion networks, which generate the overall shape of the point cloud from the incomplete point cloud and always change existing points and encounter noise and geometrical loss, PF-Net preserves the spatial arrangements of the incomplete point cloud and can figure out the detailed geometrical structure of the missing region(s) in the prediction. To succeed at this task, PF-Net estimates the missing point cloud hierarchically by utilizing a feature-points-based multi-scale generating network. Further, we add up multi-stage completion loss and adversarial loss to generate more realistic missing region(s). The adversarial loss can better tackle multiple modes in the prediction. Our experiments demonstrate the effectiveness of our method for several challenging point cloud completion tasks.

A Comprehensive Performance Evaluation for 3D Transformation Estimation Techniques

3D local feature extraction and matching is the basis for solving many tasks in the area of computer vision, such as 3D registration, modeling, recognition and retrieval. However, this process commonly draws into false correspondences, due to noise, limited features, occlusion, incomplete surface and etc. In order to estimate accurate transformation based on these corrupted correspondences, numerous transformation estimation techniques have been proposed. However, the merits, demerits and appropriate application for these methods are unclear owing to that no comprehensive evaluation for the performance of these methods has been conducted. This paper evaluates eleven state-of-the-art transformation estimation proposals on both descriptor based and synthetic correspondences. On descriptor based correspondences, several evaluation items (including the performance on different datasets, robustness to different overlap ratios and the performance of these technique combined with Iterative Closest Point (ICP), different local features and LRF/A techniques) of these methods are tested on four popular datasets acquired with different devices. On synthetic correspondences, the robustness of these methods to varying percentages of correct correspondences (PCC) is evaluated. In addition, we also evaluate the efficiencies of these methods. Finally, the merits, demerits and application guidance of these tested transformation estimation methods are summarized.

A Novel SDASS Descriptor for Fully Encoding the Information of 3D Local Surface

Local feature description is a fundamental yet challenging task in 3D computer vision. This paper proposes a novel descriptor, named Statistic of Deviation Angles on Subdivided Space (SDASS), of encoding geometrical and spatial information of local surface on Local Reference Axis (LRA). In terms of encoding geometrical information, considering that surface normals, which are usually used for encoding geometrical information of local surface, are vulnerable to various nuisances (e.g., noise, varying mesh resolutions etc.), we propose a robust geometrical attribute, called Local Minimum Axis (LMA), to replace the normals for generating the geometrical feature in our SDASS descriptor. For encoding spatial information, we use two spatial features for fully encoding the spatial information of a local surface based on LRA which usually presents high overall repeatability than Local Reference Axis (LRF). Besides, an improved LRA is proposed for increasing the robustness of our SDASS to noise and varying mesh resolutions. The performance of the SDASS descriptor is rigorously tested on four popular datasets. The results show that our descriptor has a high descriptiveness and strong robustness, and its performance outperform existing algorithms by a large margin. Finally, the proposed descriptor is applied to 3D registration. The accurate result further confirms the effectiveness of our SDASS method.