vo hoang TRONG

This work introduces a facial recognition method that aggregates weak facial parts to handle incomplete information. Using 70 facial landmarks, faces are divided into 12 parts refined via superpixel segmentation. Weak classifiers trained on these parts are combined with late-fusion voting to create a robust recognition system.

This article proposes a yielding multi-fold training (YMufT) strategy to train a DNN model on an imbalanced dataset. With the same training configurations and approximate training steps used in conventional training methods, YMufT helps the DNN model to converge faster, thus requiring less training time.

We analyze the behavior of applying a classical super-resolution (SR) method such as bicubic interpolation, and a deep CNN model such as SRCNN to enhance low-resolution (LR) weeds images used for classification. We find that SRCNN is suitable for the image size is smaller than 80 × 80.

In this paper, we study the ability of content-based image retrieval by extracting descriptors from a deep neural network (DNN) trained for classification purposes. The experiment shows that collecting features from DNN trained for weeds classification task can perform well on image retrieval.

In this study, we develop a novel classification approach via a voting method by using the late fusion of multimodal Deep Neural Networks (DNNs). The score vector used for voting is calculated by either using Bayesian conditional probability-based method or by determining priority weights.

In this paper, we use Gradient-weighted Class Activation Mapping (Grad-CAM) to visualize and analyze how well a CNN model behaves on the CNU weeds dataset.  Grad-CAM points out a CNN model can localize the object even though it is trained only for the classification problem.