Source author record

Lichao Wang

Lichao Wang appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Computer Vision Artificial Intelligence Computation and Language

Catalog footprint

What is connected

4works

3topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Beyond Perfect APIs: A Comprehensive Evaluation of LLM Agents Under Real-World API Complexity

We introduce WildAGTEval, a benchmark designed to evaluate large language model (LLM) agents' function-calling capabilities under realistic API complexity. Unlike prior work that assumes an idealized API system and disregards real-world factors such as noisy API outputs, WildAGTEval accounts for two dimensions of real-world complexity: 1. API specification, which includes detailed documentation and usage constraints, and 2. API execution, which captures runtime challenges. Consequently, WildAGTEval offers (i) an API system encompassing 60 distinct complexity scenarios that can be composed into approximately 32K test configurations, and (ii) user-agent interactions for evaluating LLM agents on these scenarios. Using WildAGTEval, we systematically assess several advanced LLMs and observe that most scenarios are challenging, with irrelevant information complexity posing the greatest difficulty and reducing the performance of strong LLMs by 27.3%. Furthermore, our qualitative analysis reveals that LLMs occasionally distort user intent merely to claim task completion, critically affecting user satisfaction.

preprint2021arXiv

The NEOLIX Open Dataset for Autonomous Driving

With the gradual maturity of 5G technology,autonomous driving technology has attracted moreand more attention among the research commu-nity. Autonomous driving vehicles rely on the co-operation of artificial intelligence, visual comput-ing, radar, monitoring equipment and GPS, whichenables computers to operate motor vehicles auto-matically and safely without human interference.However, the large-scale dataset for training andsystem evaluation is still a hot potato in the devel-opment of robust perception models. In this paper,we present the NEOLIX dataset and its applica-tions in the autonomous driving area. Our datasetincludes about 30,000 frames with point cloud la-bels, and more than 600k 3D bounding boxes withannotations. The data collection covers multipleregions, and various driving conditions, includingday, night, dawn, dusk and sunny day. In orderto label this complete dataset, we developed vari-ous tools and algorithms specified for each task tospeed up the labelling process. It is expected thatour dataset and related algorithms can support andmotivate researchers for the further developmentof autonomous driving in the field of computer vi-sion.

preprint2016arXiv

Mitosis Detection in Intestinal Crypt Images with Hough Forest and Conditional Random Fields

Intestinal enteroendocrine cells secrete hormones that are vital for the regulation of glucose metabolism but their differentiation from intestinal stem cells is not fully understood. Asymmetric stem cell divisions have been linked to intestinal stem cell homeostasis and secretory fate commitment. We monitored cell divisions using 4D live cell imaging of cultured intestinal crypts to characterize division modes by means of measurable features such as orientation or shape. A statistical analysis of these measurements requires annotation of mitosis events, which is currently a tedious and time-consuming task that has to be performed manually. To assist data processing, we developed a learning based method to automatically detect mitosis events. The method contains a dual-phase framework for joint detection of dividing cells (mothers) and their progeny (daughters). In the first phase we detect mother and daughters independently using Hough Forest whilst in the second phase we associate mother and daughters by modelling their joint probability as Conditional Random Field (CRF). The method has been evaluated on 32 movies and has achieved an AUC of 72%, which can be used in conjunction with manual correction and dramatically speed up the processing pipeline.

preprint2015arXiv

Semi-Automatic Segmentation of Autosomal Dominant Polycystic Kidneys using Random Forests

This paper presents a method for 3D segmentation of kidneys from patients with autosomal dominant polycystic kidney disease (ADPKD) and severe renal insufficiency, using computed tomography (CT) data. ADPKD severely alters the shape of the kidneys due to non-uniform formation of cysts. As a consequence, fully automatic segmentation of such kidneys is very challenging. We present a segmentation method with minimal user interaction based on a random forest classifier. One of the major novelties of the proposed approach is the usage of geodesic distance volumes as additional source of information. These volumes contain the intensity weighted distance to a manual outline of the respective kidney in only one slice (for each kidney) of the CT volume. We evaluate our method qualitatively and quantitatively on 55 CT acquisitions using ground truth annotations from clinical experts.