Search

The ability to sequentially learn multiple tasks without forgetting is a key skill of biological brains, whereas it represents a major challenge to the field of deep learning. To avoid catastrophic forgetting, various continual learning (CL) approaches have been devised. However, these usually require discrete task boundaries. This requirement seems biologically implausible and often limits the application of CL methods in the real world where tasks are not always well defined. Here, we take inspiration from neuroscience, where sparse, non-overlapping neuronal representations have been suggested to prevent catastrophic forgetting. As in the brain, we argue that these sparse representations should be chosen on the basis of feed forward (stimulus-specific) as well as top-down (context-specific) information. To implement such selective sparsity, we use a bio-plausible form of hierarchical credit assignment known as Deep Feedback Control (DFC) and combine it with a winner-take-all sparsity mechanism. In addition to sparsity, we introduce lateral recurrent connections within each layer to further protect previously learned representations. We evaluate the new sparse-recurrent version of DFC on the split-MNIST computer vision benchmark and show that only the combination of sparsity and intra-layer recurrent connections improves CL performance with respect to standard backpropagation. Our method achieves similar performance to well-known CL methods, such as Elastic Weight Consolidation and Synaptic Intelligence, without requiring information about task boundaries. Overall, we showcase the idea of adopting computational principles from the brain to derive new, task-free learning algorithms for CL. [ABSTRACT FROM AUTHOR]

Background. One of the key elements in maintaining the consistent marketing of tomato fruit is tomato quality. Since ripeness is the most important factor for tomato quality in the viewpoint of consumers, determining the stages of tomato ripeness is a fundamental industrial concern with regard to tomato production to obtain a high quality product. Since tomatoes are one of the most important crops in the world, automatic ripeness evaluation of tomatoes is a significant study topic as it may prove beneficial in ensuring an optimal production of high-quality product, increasing profitability. This article explores and categorises the various maturity/ripeness phases to propose an automated multi-class classification approach for tomato ripeness testing and evaluation. Methods. Object detection is the critical component in a wide variety of computer vision problems and applications such as manufacturing, agriculture, medicine, and autonomous driving. Due to the tomato fruits' complex identification background, texture disruption, and partial occlusion, the classic deep learning object detection approach (YOLO) has a poor rate of success in detecting tomato fruits. To figure out these issues, this article proposes an improved YOLOv5 tomato detection algorithm. The proposed algorithm CAM-YOLO uses YOLOv5 for feature extraction, target identification and Convolutional Block Attention Module (CBAM). The CBAM is added to the CAM-YOLO to focus the model on improving accuracy. Finally, non-maximum suppression and distance intersection over union (DIoU) are applied to enhance the identification of overlapping objects in the image. Results. Several images from the dataset were chosen for testing to assess the model's performance, and the detection performance of the CAM-YOLO and standard YOLOv5 models under various conditions was compared. The experimental results affirms that CAM-YOLO algorithm is efficient in detecting the overlapped and small tomatoes with an average precision of 88.1%. [ABSTRACT FROM AUTHOR]

Detection of lane region under the road boundary is an imperative module for intelligent vehicle system. Lane markings provide separate regions on the road for the vehicles to avoid the possibility of accidents. Existing methods in lane detection have limited performance using various sensor-based approaches such as Radar and LiDAR and have high operational costs. To achieve a steady and optimal lane detection, the vision-based lane region detection scheme VLDNet is proposed which utilizes a encoder-decoder network using semantic segmentation architecture. In this direction, a hybrid model using UNet and ResNet has been adopted, where UNet is used as a segmentation model and ResNet-50 is used for down-sampling the image and identifying the required features. These identified features have been then applied into UNet for up-sampling and decoding the segments of the images. The publicly available KITTI dataset have been accessed for experiments and validation of the proposed network. The method outperforms the existing state-of-the-art methods in lane region detection. The network achieves better performance using standard evaluation measures such as accuracy of 98.87%, the precision of 98.24%, recall of 96.55%, frequency weighted IoU of 97.78%, and MaxF score of 97.77%. [ABSTRACT FROM AUTHOR]