The multi-depot vehicle routing problem is a well-known non-deterministic polynomial-time hard combinatorial optimization problem, which is crucial for transportation and logistics systems. We proposed a novel fitness-scaling adaptive genetic algorithm with local search (FISAGALS). The fitness-scaling technique converts the raw fitness value to a new value that is suitable for selection. The adaptive rates strategy changes the crossover and mutation probabilities depending on the fitness value. The local search mechanism exploits the problem space in a more efficient way. The experiments employed 33 benchmark problems. Results showed the proposed FISAGALS is superior to the standard genetic algorithm, simulated annealing, tabu search, and particle swarm optimization in terms of success instances and computation time. Furthermore, FISAGALS performs better than parallel iterated tabu search (PITS) and fuzzy logic guided genetic algorithm (FLGA), and marginally worse than ILS-RVND-SP in terms of the maximum gap. It performs faster than PITS and ILS-RVND-SP (a combination of iterated local search framework [ILS], a variable neighborhood descent with random neighborhood ordering [RVND] and the the set partitioning [SP] model) and slower than FLGA. In summary, FISAGALS is a competitive method with state-of-the-art algorithms.

No abstract

Scribble labels have gained increasing attention in the field of weakly supervised video salient object detection (VSOD). Based on scribble labels, latest methods can spread labeled pixels to unlabeled regions using local coherence loss, but predicted objects often lose detail and boundary information. In this work, a novel method based on back‐foreground weight contrast is proposed that adds label enhancement points to facilitate the model to learn the edge, detail and location of salient object. Additionally, a new VSOD framework based on global structural localization is introduced. Enhanced scribble labels are used to assist the model for global localization, and then the located regions are finely segmented by the trained model. Extensive experiments demonstrate that the method achieves the state‐of‐the‐art performance on common VSOD datasets, with an improvement of 3.75%, 4.68%, and 0.88% in S‐measure, F‐measure, and MAE, respectively.