This project is supported by King’s Together: Multi & Interdisciplinary Research Scheme which brings expertise from both Informatics and Geography to work on developing new machine learning approaches linked with (new) remote sensing data sources.

Our focus will be to develop algorithms to separate dammed reservoirs from the millions of other water bodies on Earth. Pilot studies that we have carried out in the Volta and Limpopo basins indicate that dammed reservoirs tend to have a triangular or elongate triangular shape, whereas natural reservoirs tend to be round.

The goal is to employ machine learning techniques, e.g. deep neural networks, to detect and recognize dammed reservoirs in remote sensing imagery. By identifying water bodies spectrally in earth observation imagery and then analysing their shape we hope to extend our database of large and medium sized dams to the many millions of small dams important for smallholder irrigation.

Using transfer learning to understand visual objects and their relationships

Machine Perception tasks have flourished since the advent of deep learning techniques. Next key problem lies on visual scene understanding. To make sense of visual scenes, we need to rely on the visual object relationships inside, e.g. person riding a bike, book on the table. The challenge for this task is that 1) the training data is limited, on particular those unusual seen objects/object relationships; 2) visual relationships become complicated and diverse with an increase of object numbers. Recent works are mainly focused on detecting the normal relationships between any two objects.

This research shall employ the transfer learning methods to transfer available knowledge of visual relationships to new objects with unknown relationships. The significance of this research is not just to enhance the machine perception ability. The largest public dataset with full annotations (e.g. pixel classes/object bounding boxes) contains several million images, while billions of images on the Internet have no labels or only image-level labels. Research on transfer learning allows us to leverage a relatively small amount of expensively annotated images to detect new objects and their relationships in a much larger dataset without or with only cheap image-level labels. This is particularly important in the information era.

Open-Set Surgical Instrument Segmentation with Endoscopic Vision-Language Model

Automatic segmentation of instruments from laparoscopic surgery images or videos plays a key role in providing advanced assistance to the clinical team, with emerging applications in enhancing surgery conducted by humans and ultimately in semi-autonomous robot-human delivered surgery and surgical imaging techniques. The state-of-the-art solution for it is to train deep convolutional networks from manually annotated datasets. Annotating such datasets from surgical videos is however tedious and time-consuming.

In order to achieve high robustness and accuracy of surgical segmentation, it is critical to make use of large amount of non-annotated data in the open-set (real world). Second, owing to the evolutive nature of surgical technology, the segmentation concepts (classes) actually increase, the open-set contains data with both base and new classes; in contrast, trained surgical instrument detectors can only detect base classes in the curated set. How to equip such detectors trained on base classes in a curated set with the ability to detect new classes in the open-set becomes another critical problem.

This project aims at overcoming the limitations of open-set surgical segmentation with the help of an endoscopic vision-language model. Vision-language model is a recently developed technique proved to be highly powerful to open-set visual recognition. It is originally pretrained on millions of natural image-text pairs over tremendous classes and is established mainly for image classification task. To use this technique in the medical imaging domain, we propose to train a new endoscopic vision-language model dedicated to robotic surgery given the access to the Cambridge-1. The model will be made public for the broader community to make use of it for many downstream autonomous endoscopic tasks.

This project will utilize machine learning techniques to analyse the crowd densities in the Strand /Aldwych region (SAR), which is currently under a major redevelopment scheme by Westminster city council (WCC). The output of this project aligns with the aims of this redevelopment scheme: being able to know the crowd counts in different areas and periods of SAR would inform how visitors use different spaces within SAR and therefore help WCC improve the public realm to provide a better pedestrian experience.

The crowd movement in the crossroads will particularly help make more informed decisions around personal safety by reducing traffic in some areas while mitigating the effects of traffic in other areas, which shall lead to better movement of the traffic as well as improve the air quality in SAR. Special attentions will be paid to the crowd movement in the bike and pedestrian lanes to improve the links for waling and cycling.

Finally, this project will also support and enhance the area's economy by analysing the dwell time of crowds in different designed spaces and therefore guide future investment initiatives. Overall, in a sustainable viewpoint, the techniques and analysis methods developed in this project will allow WCC to regularly monitor the change of crowd densities and movements in SAR, hence advise the investment and

The China Postdoctoral Science Foundation Project (CPSF) is a national-level initiative aimed at supporting postdoctoral researchers in China. Established by the Ministry of Human Resources and Social Security, the CPSF provides funding to eligible postdoctoral researchers across various fields of science and technology.

The project aims to foster scientific research and innovation by offering financial support, resources, and opportunities for postdoctoral fellows to conduct research, publish papers, collaborate with academic institutions, and contribute to the advancement of science and technology in China.

Shanghai Super Postdoc Award is an award launched by the Shanghai municipal government to attract and cultivate high-level talents. This initiative aims to provide a strong research platform and excellent working environment for postdoctoral researchers, enabling them to achieve significant research outcomes and make breakthroughs in academic, technological, and innovative domains.

Applicants not only have access to substantial research funding and academic support but also the opportunity to collaborate with outstanding research teams and scientific institutions.