Exploring better target for shadow detection

Handwritten Dongba character recognition (HDCR) is a challenging task due to different recognition granularity and it plays an important role in history and philology. Recently, deep learning-based methods are adopted and outperform traditional approaches. However, they do not well explore the multi-granularity information, leading to many highly similar Dongba characters cannot be recognized accurately. In this paper, a Multiple Attentional Aggregation Network (MAAN) is proposed for HDCR. Specifically, a novel Hybrid Attentional Mapping Unit (HAMU) is designed for multi-scale and richer feature expression. The hybrid attention mechanism in HAMU can enhance the discrimination between important and redundant features. Besides, a novel Spatial Attentional Aggregation Unit (SAAU) is adopted to aggregate multi-scale features of different depths. In this way, more details from large scale features and global semantic information from small scale features will be adequately aggregated. Experiments are conducted on our constructed novel Dongba character image dataset and validate the effectiveness and superiority of our methods and the extensive experimental results demonstrate that our proposed MAAN outperforms state-of-the-art algorithms. 

Multiple attentional aggregation network for handwritten Dongba character recognition

With the widespread use of touch-screen devices, it is more and more convenient for people to draw sketches on screen. This results in the demand for automatically understanding the sketches. Thus, the sketch recognition task becomes more significant than before. To accomplish this task, it is necessary to solve the critical issue of improving the distinction of the sketch features. To this end, we have made efforts in three aspects. First, a novel multi-scale residual block is designed. Compared with the conventional basic residual block, it can better perceive multi-scale information and reduce the number of parameters during training. Second, a hierarchical residual structure is built by stacking multi-scale residual blocks in a specific way. In contrast with the single-level residual structure, the learned features from this structure are more sufficient. Last but not least, the compact triplet-center loss is proposed specifically for the sketch recognition task. It can solve the problem that the triplet-center loss does not fully consider too large intra-class space and too small inter-class space in sketch field. By studying the above modules, a hierarchical residual network as a whole is proposed for sketch recognition and evaluated on Tu-Berlin benchmark thoroughly. The experimental results show that the proposed network outperforms most of baseline methods and it is excellent among non-sequential models at present. 

/pdf/a-hierarchical-residual-network-with-compact-triplet-center-salfxc9n.pdf

A hierarchical residual network with compact triplet-center loss for sketch recognition

With the widespread use of touch-screen devices, it is more and more convenient for people to draw sketches on screen. This results in the demand for automatically understanding the sketches. Thus, the sketch recognition task becomes more significant than before. To accomplish this task, it is necessary to solve the critical issue of improving the distinction of the sketch features. To this end, we have made efforts in three aspects. First, a novel multi-scale residual block is designed. Compared with the conventional basic residual block, it can better perceive multi-scale information and reduce the number of parameters during training. Second, a hierarchical residual structure is built by stacking multi-scale residual blocks in a specific way. In contrast with the single-level residual structure, the learned features from this structure are more sufficient. Last but not least, the compact triplet-center loss is proposed specifically for the sketch recognition task. It can solve the problem that the triplet-center loss does not fully consider too large intra-class space and too small inter-class space in sketch field. By studying the above modules, a hierarchical residual network as a whole is proposed for sketch recognition and evaluated on Tu-Berlin benchmark thoroughly. The experimental results show that the proposed network outperforms most of baseline methods and it is excellent among non-sequential models at present.

A hierarchical residual network with compact triplet-center loss for sketch recognition.

Shadow removal is a fundamental and pivotal task to build the high-level cognition in the computer vision field. Due to the fact that the existing shadow removal methods cannot effectively remove shadows from the outdoor image and deal with shadow boundaries, we construct a convolutional neural network without the process of shadow detection for the shadow removal task. The constructed CNN avoids the risk of gradient vanishing by designing double-attention residual block and improves the performance of shadow removal by fusing the knowledge transfer idea. Specially, we design a hand-crafted feature, named brightness-gradient difference feature, to distinguish shadow boundary pixels from non-shadow boundary pixels, and the designed feature is fused into the loss function to dilute or even eliminate the existing shadow boundaries. Extensive experiments using three public shadow removal benchmarks with three measurable indicators are reported in this paper. The results of experiments demonstrate that the proposed method has an effective performance for the shadow removal task. The ablation studies validate the structural rationality of the proposed method. 

Botanical Farms CBD Gummies

Papers

A hierarchical residual network with compact triplet-center loss for sketch recognition

A hierarchical residual network with compact triplet-center loss for sketch recognition.

Brightness-gradient difference feature guided shadow removal method