Acquiring the most representative examples via active learning (AL) can benefit many data-dependent computer vision tasks by minimizing efforts of image-level or pixel-wise annotations. In this paper, we propose a novel Collaborative Panoptic-Regional Active Learning framework (CPRAL) to address the semantic segmentation task. For a small batch of images initially sampled with pixel-wise annotations, we employ panoptic information to initially select unlabeled samples. Considering the class imbalance in the segmentation dataset, we import a Regional Gaussian Attention module (RGA) to achieve semantics-biased selection. The subset is highlighted by vote entropy and then attended by Gaussian kernels to maximize the biased regions. We also propose a Contextual Labels Extension (CLE) to boost regional annotations with contextual attention guidance. With the collaboration of semantics-agnostic panoptic matching and region-biased selection and extension, our CPRAL can strike a balance between labeling efforts and performance and compromise the semantics distribution. We perform extensive experiments on Cityscapes and BDD10K datasets and show that CPRAL outperforms the cutting-edge methods with impressive results and less labeling proportion.

Nitrogen (N) is an essential macronutrient for plant function and growth and a key component of amino acids, which form the building blocks of plant proteins and enzymes. However, misuse and overuse of N can have many negative impacts on the ecosystem, such as reducing soil exchangeable base cations (BCs) and causing soil acidification. In this research, we evaluated clonal Chinese fir (Cunninghamia lanceolata (Lamb.) Hook) seedlings grown with exponentially increasing N fertilization (0, 0.5, 1, 2 g N seedling−1) for a 100-day trial in a greenhouse. The growth of seedlings, their nutrient contents, and soil exchangeable cations were measured. We found that N addition significantly increased plant growth and N content but decreased phosphorous (P) and potassium (K) contents in plant seedlings. The high nitrogen (2 g N seedling−1) treated seedlings showed a negative effect on growth, indicating that excessive nitrogen application caused damage to the seedlings. Soil pH, soil exchangeable base cations (BCs), soil total exchangeable bases (TEB), soil cation exchange capacity (CEC), and soil base saturation (BS) significantly decreased following N application. Our results implied that exponential fertilization resulted in soil acidification and degradation of soil capacity for supplying nutrient cations to the soil solution for plant uptake. In addition, the analysis of plants and BCs revealed that Na+ is an important base cation for BCs and for plant growth in nitrogen-induced acidified soils. Our results provide scientific insights for nitrogen application in seedling cultivation in soils and for further studies on the relationship between BCs and plant growth to result in high-quality seedlings while minimizing fertilizer input and mitigating potential soil pollution.

Acquiring the most representative examples via active learning (AL) can benefit many data-dependent computer vision tasks by minimizing efforts of image-level or pixel-wise annotations. In this paper, we propose a novel Collaborative Panoptic-Regional Active Learning framework (CPRAL) to address the semantic segmentation task. For a small batch of images initially sampled with pixel-wise annotations, we employ panoptic information to initially select unlabeled samples. Considering the class imbalance in the segmentation dataset, we import a Regional Gaussian Attention module (RGA) to achieve semantics-biased selection. The subset is highlighted by vote entropy and then attended by Gaussian kernels to maximize the biased regions. We also propose a Contextual Labels Extension (CLE) to boost regional annotations with contextual attention guidance. With the collaboration of semantics-agnostic panoptic matching and regionbiased selection and extension, our CPRAL can strike a balance between labeling efforts and performance and compromise the semantics distribution. We perform extensive experiments on Cityscapes and BDD10K datasets and show that CPRAL outperforms the cutting-edge methods with impressive results and less labeling proportion.