Video-Assisted Thoracoscopic Surgery (VATS) is a minimally invasive procedure developed to remove specific lung segments for the treatment of early-stage lung diseases. The surgical procedure involves intricate vascular and bronchial anatomy to preserve as much lung tissue as possible, minimizing impact on the pulmonary function. To assist in monitoring and early warning of this high-risk surgical workflow, we build a new dataset, LungRes80, including 269,806 video frames with phase annotations sampled from 80 VATS cases. LungRes80 presents unique challenges for hierarchical temporal modeling due to diverse short-term transitions between segmentectomy phases and latent long-term causal relations. To this end, we introduce an online baseline model termed LungReco. This framework employs Masked Causal Reasoning (MCR) to perform causal reasoning with semantic modeling from continuously updated memories along with pre-trained Large Language Models (LLMs), and combines it with Concurrent Spatial-Temporal encoding (CoST) for holistic bi-modal co-spatial-temporal aggregation across short- and long-term memories. Furthermore, a new metric, called the Attentional Distraction Coefficient (ADC), is proposed to quantify the costs of intraoperative distraction and postoperative corrections by wrong predictions. We establish a comprehensive benchmark for surgical workflow recognition by evaluating representative models on LungRes80, AutoLaparo, and Cholec80, where our method consistently achieves state-of-the-art performance. Code and data are available at LungRes80.

We present a novel framework, CLIP-SP, and a novel adaptive prompt method to leverage pre-trained knowledge from CLIP for scene parsing. Our approach addresses the limitations of DenseCLIP, which demonstrates the superior image segmentation provided by CLIP pre-trained models over ImageNet pre-trained models, but struggles with rough pixel-text score maps for complex scene parsing. We argue that, as they contain all textual information in a dataset, the pixel-text score maps, i.e., dense prompts, are inevitably mixed with noise. To overcome this challenge, we propose a two-step method. Firstly, we extract visual and language features and perform multi-label classification to identify the most likely categories in the input images. Secondly, based on the top-k categories and confidence scores, our method generates scene tokens which can be treated as adaptive prompts for implicit modeling of scenes, and incorporates them into the visual features fed into the decoder for segmentation. Our method imposes a constraint on prompts and suppresses the probability of irrelevant categories appearing in the scene parsing results. Our method achieves competitive performance, limited by the available visual-language pre-trained models. Our CLIP-SP performs 1.14% better (in terms of mIoU) than DenseCLIP on ADE20K, using a ResNet-50 backbone.

Sea fog detection is a challenging and significant task in the field of remote sensing. Deep learning-based methods have shown promising potential, but require a large amount of pixel-level labeled data that are time-consuming and labor-intensive to acquire. To scale up the dataset and overcome the limitations of pixel-level annotation, we attempt to explore the existing knowledge from historical statistics for label efficient sea fog detection. In this paper, we propose an image-level Weakly Supervised Sea Fog Detection Dataset (WS-SFDD) and a novel weakly supervised sea fog detection framework via prototype learning, named ProCAM. According to the sea fog events recorded by the Marine Weather Review published quarterly by the National Meteorological Center of China, we collect the sea fog images from Himawari-8 satellite data and obtain free image-level labels to construct the dataset. However, with image-level annotations, existing weakly supervised semantic segmentation methods mainly rely on class activation maps (CAMs) and have limitations when applied to such a specific scenario: 1) the pseudo labels mainly cover the most discriminative part of object regions that are incomplete; 2) the background is complex with varying atmospheric conditions and it is difficult to distinguish sea fog from low clouds due to their high similarity in spectral characteristics; 3) the co-occurring context like ‘sea’ distracts the model and thus degrades the performance. To address the above issues, in our proposed ProCAM, we first design a prototype re-activation (PRA) module that reactivates self-similar sea fog regions by pixel-to-prototype feature matching to improve the robustness and completeness of CAMs. Then, we develop a pixel-to-prototype contrastive (PPC) learning method to increase the distance between sea fog and background in the embedding space for learning more discriminative dense features. Finally, a self-augmented regularization (SAR) strategy is presented to decouple sea fog from its co-occurring context and thus avoid background interference. Extensive experiments on the WS-SFDD dataset demonstrate our proposed method ProCAM achieves superior performance with an F1-score of 77.59% and a critical success index of 63.39%. To the best of our knowledge, this is the first work to perform image-level weakly supervised sea fog detection in remote sensing images. The dataset and code are available at https://github.com/yixianghuang/ProCAM.

For integers $k$ and $d$ with $k\ge 2d>0$, a circular ${k}/{d}$-flow of a graph $G$ is an orientation together with a mapping from $E(G)$ to $\{\pm d, \pm (d+1),\ldots,\pm (k-d)\}$ such that, for each vertex of $G$, the sum of images on outgoing edges is equal to the sum of images on incoming edges. Related to the four color problem, a classical result of Tutte shows that a cubic graph admits a circular $4/1$-flow if and only if it is Class I (i.e., $3$-edge-colorable). Tutte's $3$-flow conjecture implies that every $5$-regular Class I graph admits a nowhere-zero $3$-flow (equivalently, a circular $6/2$-flow) as a special case. Steffen in 2015 conjectured that every $(2t+1)$-regular Class I graph admits a circular $(2t+2)/t$-flow. He also proposed a more general conjecture that every $(2t+1)$-odd-edge-connected $(2t+1)$-regular graph admits a circular $(2t+2)/t$-flow for any integer $t\ge 2$, which includes the circular flow conjecture of Jaeger (1981) stating that every $2t$-edge-connected graph admits a circular $(2t+2)/t$-flow for any even $t\ge 2$. Jaeger's conjecture was disproved in 2018 for all even $t\ge 6$, and based on these results, Mattiolo and Steffen recently constructed counterexamples to Steffen's conjecture for Class I graphs when $t=4k+2$ for any integer $k\ge 1$. In this paper, we extend the above results and construct infinitely many $2t$-edge-connected $(2t+1)$-regular Class I graphs without circular $(2t+2)/t$-flows for any integer $t\in \{6,8,10\}$ or $t\geq 12$. Our result provides more general counterexamples to Steffen's two conjectures for both even and odd $t$ and simultaneously generalizes the counterexamples of Jaeger's circular flow conjecture to regular Class I graphs.

This paper studies the fundamental relations among integer flows, modulo orientations, integer-valued and real-valued circular flows, and monotonicity of flows in signed graphs. A (signed) graph is modulo-$(2p+1)$-orientable if it has an orientation such that the indegree is congruent to the outdegree modulo $2p+1$ at each vertex. An integer-valued $\frac{2p+1}{p}$-flow is a flow taking integer values in $\{\pm p, \pm (p+1)\}$. Extending a fundamental result of Jaeger to signed graphs, we show that a bridgeless signed graph is modulo-$(2p+1)$-orientable if and only if it admits an integer-valued $\frac{2p+1}{p}$-flow. It was conjectured by Raspaud and Zhu that, for any signed graph, the admission of a circular $r$-flow implies the admission of an integer-valued $\lceil r \rceil$-flow. Although this conjecture has been disproved in general, it is confirmed in this paper for bridgeless signed graphs if $r=\frac{2p+1}{p}$ and $p \geq 3$.

For integers $a\ge 2b>0$, a circular $a/b$-flow is a flow that takes values from $\{\pm b, \pm(b+1), \dots, \pm(a-b)\}$. The Planar Circular Flow Conjecture states that every $2k$-edge-connected planar graph admits a circular $(2+\frac{2}{k})$-flow. The cases $k=1$ and $k=2$ are equivalent to the Four Color Theorem and Grötzsch's 3-Color Theorem. For $k\ge 3$, the conjecture remains open. Here we make progress when $k=4$ and $k=6$. We prove that (i) every 10-edge-connected planar graph admits a circular $5/2$-flow and (ii) every 16-edge-connected planar graph admits a circular 7/3-flow. The dual version of statement (i) on circular coloring was previously proved by Dvořák and Postle [ Combinatorica, 37 (2017), pp. 863--886], but our proof has the advantages of being much shorter and avoiding the use of computers for case-checking. Further, it has new implications for antisymmetric flows. Statement (ii) is especially interesting because the counterexamples to Jaeger's original Circular Flow Conjecture are 12-edge-connected nonplanar graphs that admit no circular 7/3-flow. Thus, the planarity hypothesis of (ii) is essential.

The anti-Ramsey problem was introduced by Erdös, Simonovits, and Sós in 1970s. The anti-Ramsey number of a hypergraph H, ar(n,s, H), is the smallest integer c such that in any coloring of the edges of the s-uniform complete hypergraph on n vertices with exactly c colors, there is a copy of H whose edges have distinct colors. In this paper, we determine the anti-Ramsey numbers of linear paths and loose paths in hypergraphs for sufficiently large n and give bounds for the anti-Ramsey numbers of Berge paths. Similar exact anti-Ramsey numbers are obtained for linear/loose cycles, and bounds are obtained for Berge cycles. Our main tools are the path extension technique and stability results on hypergraph Turán problems of paths and cycles.

A mod $(2p+1)$-orientation $D$ is an orientation of $G$ such that $d_D^+(v)-d_D^-(v)\equiv 0 \pmod {2p+1}$ for any vertex $v \in V(G)$. Jaeger conjectured that every $4p$-edge-connected graph has a mod $(2p+1)$-orientation. A graph $G$ is strongly ${\mathbb Z}_{2p+1}$-connected if for every mapping $b: V(G) \mapsto {\mathbb Z}_{2p+1}$ with $\sum_{v\in V(G)}b(v)=0$, there exists an orientation $D$ of $G$ such that $d_D^+(v)-d_D^-(v)= b(v)$ in ${\mathbb Z}_{2p+1}$ for any $v \in V(G)$. A strongly ${\mathbb Z}_{2p+1}$-connected graph admits a mod $(2p+1)$-orientation, and it is a contractible configuration for mod $(2p+1)$-orientation. We prove Jaeger's module orientation conjecture is equivalent to its restriction to bipartite simple graphs and investigate strongly ${\mathbb Z}_{2p+1}$-connectedness of certain bipartite graphs, particularly for $p=2$. We also show that if $G$ is a simple graph with $|V(G)|\ge N(p)= 1152p^4$ and $\min\{\delta(G),\delta(G^c)\}\ge 4p$, then either $G$ or $G^c$ is strongly ${\mathbb Z}_{2p+1}$-connected. When $p=2$, the value of $N(2)$ can be reduced to $N(2) = 80$.

Let $\mathbb Z_m$ be the cyclic group of order $m \geq 3$. A graph $G$ is $\mathbb Z_m$-connected if $G$ has an orientation $D$ such that for any mapping $b: V(G) \mapsto \mathbb Z_m$ with $\sum_{v\in V(G)}b(v)=0$, there exists a mapping $f:E(G) \mapsto \mathbb Z_m -\{0\}$ satisfying $\sum_{e\in E_D^+(v)} f(e) - \sum_{e\in E_D^-(v)} f(e) = b(v)$ in $\mathbb Z_m$ for any $v \in V(G)$; and a graph $G$ is strongly $\mathbb Z_m$-connected if, for any mapping $\theta: V(G)\rightarrow \mathbb Z_m$ with $\sum_{v\in V(G)}\theta(v) = |E(G)|$ in $\mathbb Z_m$, there is an orientation $D$ such that $d_D^+(v)=\theta(v)$ in $\mathbb Z_m$ for each $v \in V(G)$. In this paper, we study the relation between $\mathbb Z_m$-connected graphs and strongly $\mathbb Z_m$-connected graphs and show that a graph $G$ is $\mathbb Z_m$-connected if and only if $(m-2)G$ is strongly $\mathbb Z_m$-connected, where $(m-2)G$ is the graph obtained from $G$ by replacing each edge in $G$ with $m-2$ parallel edges. We also show that if $G$ is $\mathbb Z_m$-connected, then $(m-2)G$ has $m-1$ edge disjoint spanning trees. Those results together with a result by Jaeger et al. [J. Combin. Theory Ser. B, 56 (1992), pp. 165-182] imply that every $\mathbb Z_3$-connected graph is $A$-connected for any abelian group $A$ with $|A| \geq 4$. They are applied to determine the exact values of $ex(n,\mathbb Z_m)$ for all $m\geq 3$, where $ex(n,\mathbb Z_m)$ is the largest integer such that every simple graph on $n$ vertices with at most $ex(n,\mathbb Z_m)$ edges is not $\mathbb Z_m$-connected, and to present characterizations of graphic and multigraphic sequences that have $\mathbb Z_m$-connected realizations.