Quality inspection is inevitable in the steel industry so there are already benchmark datasets for the visual inspection of steel surface defects. In our work, we show, contrary to previous recent articles, that a generic state-of-art deep neural network is capable of almost-perfect classification of defects of two popular benchmark datasets. However, in real-life applications new types of errors can always appear, thus incremental learning, based on very few example shots, is challenging. In our article, we address the problems of the low number of available shots of new classes, the catastrophic forgetting of known information when tuning for new artifacts, and the long training time required for re-training or fine-tuning existing models. In the proposed new architecture we combine EfficientNet deep neural networks with randomized classifiers to aim for an efficient solution for these demanding problems. The classification outperforms all other known approaches, with an accuracy 100% or almost 100%, on the two datasets with the off-the-shelf network. The proposed few-shot learning approach shows considerably higher accuracy at a low number of shots than the different methods under testing, while its speed is significantly (at least 10 times) higher than its competitors. According to these results, the classification and few-shot learning of steel surface defects can be solved more efficiently than was possible before.
Particle filter has grown to be a standard tool for solving visual tracking problems in real world applications. One of the critical tasks in object tracking is the tracking of fast-moving objects in complex environments, which contain cluttered background and scale change. In this paper, a new tracking algorithm is presented by using the joint color texture histogram to represent a target and then applying it to particle filter algorithm called PFJCTH. The texture features of the object are extracted by using the local binary pattern (LBP) technique to represent the object. The proposed algorithm extracts effectively the edge and corner features in the target region, which characterize better and represent more robustly the target. The experiments showed that this new proposed algorithm produces excellent tracking results and outperforms other tracking algorithms.
31 July 2020 in Signal, Image and Video Processing
Our paper deals with active multiview object recognition focusing on the directional support of sequential multiple shots. Since inertial sensors are easily available nowadays, we propose the use of them to estimate the orientation change of the camera and thus to estimate the probability of relative poses. With the help of relative orientation change, we can compute transition probabilities between possible poses and can use a hidden Markov model to evaluate state (pose) sequences and can thus increase the recognition rate. Furthermore, we can plan our next viewing position to minimize the risk of misclassification, resulting in higher overall recognition rates. Besides giving the theoretical details, we use two datasets to illustrate the performance of our model through several tests including occlusion, blur, Gaussian noise, and to compare to a solution with a long short-term memory network.
Pill image detection is essential in healthcare applications such as hospital workflow optimization, assistance for visually impaired individuals, and elderly care. However, high inter-class similarity and varying imaging conditions make accurate recognition challenging. This paper presents UMobile-Net, a dual-model deep learning framework integrating a segmentation model (U-Net or W 2) with a MobileNet-based classifier in a two-stage pipeline. In stage one, segmentation masks derived from dataset annotations are refined to extract pill regions, followed by image enhancement. In stage two, the processed images are classified using a customized MobileNet architecture. Evaluations on the CURE and OGYEI-v2 datasets show that UMobile-Net consistently surpasses existing methods. The U-Net–based variant achieved accuracies of 97.78% (CURE) and 97.62% (OGYEI-v2), while the W 2 –based variant achieved 95.85% and 97.61%, respectively. The results confirm the robustness and accuracy of the proposed approach for real-world pill recognition.