We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to asemantic cluster formed by local decoder feature points with semantic similarity. By rethinking point affiliation, we present a generic formulation for generating upsampling kernels. The kernels encourage notonly semantic smoothness butalsoboundary sharpness intheupsampled feature maps. Such properties are particularly useful for some dense prediction tasks such as semantic segmentation. The key idea of our formulation istogenerate similarity-awarekernels bycomparing thesimilarity between each encoder feature point and the spatially associated local region of decoder features.

We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to a semantic cluster formed by local decoder feature points with semantic similarity. By rethinking point affiliation, we present a generic formulation for generating upsampling kernels. The kernels encourage not only semantic smoothness but also boundary sharpness in the upsampled feature maps. Such properties are particularly useful for some dense prediction tasks such as semantic segmentation. The key idea of our formulation is to generate similarity-aware kernels by comparing the similarity between each encoder feature point and the spatially associated local region of decoder features. In this way, the encoder feature point can function as a cue to inform the semantic cluster of upsampled feature points. To embody the formulation, we further instantiate a lightweight upsampling operator, termed Similarity-Aware Point Affiliation (SAPA), and investigate its variants. SAPA invites consistent performance improvements on a number of dense prediction tasks, including semantic segmentation, object detection, depth estimation, and image matting. Code is available at: https://github.com/poppinace/sapa

We first visualize the qualitative results of different upsampling operators. Semantic segmentation visualizations are shown in Fig. S1, image matting ones are shown in Fig. S2, To better understand how SAP A works, we supplement additional visualizations on the encoder features, decoder features, upsampling kernels, and upsampled features of SAP A. As shown in Fig. S4, for the upsampling kernels, we choose every top-left weight of the upsampling kernel for visualization, therefore the kernel map is of the same size with the upsampled feature. The backbone VGG-16 is pretrained on ImageNet. Semantic Segmentation on ADE20K We use the codes released by the authors. We keep all other settings unchanged while only modify the upsampling stages.

We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to a semantic cluster formed by local decoder feature points with semantic similarity. By rethinking point affiliation, we present a generic formulation for generating upsampling kernels. The kernels encourage not only semantic smoothness but also boundary sharpness in the upsampled feature maps. Such properties are particularly useful for some dense prediction tasks such as semantic segmentation. The key idea of our formulation is to generate similarity-aware kernels by comparing the similarity between each encoder feature point and the spatially associated local region of decoder features.

Many real world planning problems require goals with deadlines anddurative actions that consume resources. In this paper, we present Sapa, a domain-independent heuristic forward chaining planner thatcan handle durative actions, metric resource constraints, and deadlinegoals. The main innovation of Sapa is the set of distance basedheuristics it employs to control its search. We consider bothoptimizing and satisficing search. For the former, we identifyadmissible heuristics for objective functions based on makespan andslack. For satisficing search, our heuristics are aimed at scalabilitywith reasonable plan quality. Our heuristics are derived from the relaxed temporal planning graph'' structure, which is ageneralization of planning graphs to temporal domains. We also providetechniques for adjusting the heuristic values to account for resourceconstraints. Our experimental results indicate that Sapa returnsgood quality solutions for complex planning problems in reasonabletime.

Many real world planning problems require goals with deadlines anddurative actions that consume resources. In this paper, we present Sapa, a domain-independent heuristic forward chaining planner thatcan handle durative actions, metric resource constraints, and deadlinegoals. The main innovation of Sapa is the set of distance basedheuristics it employs to control its search. We consider bothoptimizing and satisficing search. For the former, we identifyadmissible heuristics for objective functions based on makespan andslack. For satisficing search, our heuristics are aimed at scalabilitywith reasonable plan quality. Our heuristics are derived from the``relaxed temporal planning graph'' structure, which is ageneralization of planning graphs to temporal domains. We also providetechniques for adjusting the heuristic values to account for resourceconstraints. Our experimental results indicate that Sapa returnsgood quality solutions for complex planning problems in reasonabletime.

The success of the Deep Space Remote Agent experiment has demonstrated the promise and importance of metric temporal planning for real-world applications. HSTS/RAX, the planner used in the remote agent experiment, was predicated on the availability of domain-and planner-dependent control knowledge, the collection and maintenance of which is admittedly a laborious and errorprone activity. An obvious question is whether it will be possible to develop domain-independent metric temporal planners that are capable of scaling up to such domains. The past experience has not been particularly encouraging. Although there have been some ambitious attempts-including IxTeT (Ghallab & Laruelle, 1994) and Zeno (Penberthy & Well, 1994), their performance has not been particularly satisfactory. Some encouraging signs however are the recent successes of domain-independent heuristic planning techniques in classical planning (c.f., Nguyen, Kambhampati, & Nigenda, 2001; Bonet, Loerincs, & Geffner, 1997; Hoffmann & Nebel, 2001). Our research is aimed at building on these successes to develop a scalable metric temporal planner. At first blush search control for metric temporal planners would seem to be a very simple matter of adapting the work on heuristic planners in classical planning (Bonet et al., 1997; Nguyen et al., 2001; Hoffmann & Nebel, 2001). The adaptation however does pose several challenges: - Metric temporal planners tend to have significantly larger search spaces than classical planners.

Sapa is a domain-independent heuristic forward chaining planner that can handle durative actions, metric resource constraints, and deadline goals. It is designed to be capable of handling the multi-objective nature of metric temporal planning. Our technical contributions include (i) planning-graph based methods for deriving heuristics that are sensitive to both cost and makespan (ii) techniques for adjusting the heuristic estimates to take action interactions and metric resource limitations into account and (iii) a linear time greedy post-processing technique to improve execution flexibility of the solution plans. An implementation of Sapa using many of the techniques presented in this paper was one of the best domain independent planners for domains with metric and temporal constraints in the third International Planning Competition, held at AIPS-02. We describe the technical details of extracting the heuristics and present an empirical evaluation of the current implementation of Sapa.