Manhattan-distance IOU loss for fast and accurate bounding box regression and object detection

A Multi-Scale Target Detection Method Using an Improved Faster Region Convolutional Neural Network Based on Enhanced Backbone and Optimized Mechanisms

Currently, existing deep learning methods exhibit many limitations in multi-target detection, such as low accuracy and high rates of false detection and missed detections. This paper proposes an improved Faster R-CNN algorithm, aiming to enhance the algorithm's capability in detecting multi-scale targets. This algorithm has three improvements based on Faster R-CNN. Firstly, the new algorithm uses the ResNet101 network for feature extraction of the detection image, which achieves stronger feature extraction capabilities. Secondly, the new algorithm integrates Online Hard Example Mining (OHEM), Soft non-maximum suppression (Soft-NMS), and Distance Intersection Over Union (DIOU) modules, which improves the positive and negative sample imbalance and the problem of small targets being easily missed during model training. Finally, the Region Proposal Network (RPN) is simplified to achieve a faster detection speed and a lower miss rate. The multi-scale training (MST) strategy is also used to train the improved Faster R-CNN to achieve a balance between detection accuracy and efficiency. Compared to the other detection models, the improved Faster R-CNN demonstrates significant advantages in terms of mAP@0.5, F1-score, and Log average miss rate (LAMR). The model proposed in this paper provides valuable insights and inspiration for many fields, such as smart agriculture, medical diagnosis, and face recognition.

An improved algorithm based on YOLOv5 for detecting Ambrosia trifida in UAV images

A YOLOv5-based YOLOv5-KE unmanned aerial vehicle (UAV) image detection algorithm is proposed to address the low detection accuracy caused by the small size, high density, and overlapping leaves of Ambrosia trifida targets in UAV images. The YOLOv5-KE algorithm builds upon the YOLOv5 algorithm by adding a micro-scale detection layer, adjusting the hierarchical detection settings based on k-Means for Anchor Box, improving the loss function of CIoU, reselecting and improving the detection box fusion algorithm. Comparative validation experiments of the YOLOv5-KE algorithm for Ambrosia trifida recognition were conducted using a self-built dataset. The experimental results show that the best detection accuracy of Ambrosia trifida in UAV images is 93.9%, which is 15.2% higher than the original YOLOv5. Furthermore, this algorithm also outperforms other existing object detection algorithms such as YOLOv7, DC-YOLOv8, YOLO-NAS, RT-DETR, Faster RCNN, SSD, and Retina Net. Therefore, YOLOv5-KE is a practical algorithm for detecting Ambrosia trifida under complex field conditions. This algorithm shows good potential in detecting weeds of small, high-density, and overlapping leafy targets in UAV images, it could provide technical reference for the detection of similar plants.

Fish Physiologically Based Toxicokinetic Modeling Approach for In Vitro-In Vivo and Cross-Species Extrapolation of Endocrine-Disrupting Chemicals in Risk Assessment

High-throughput in vitro assays combined with in vitro-in vivo extrapolation (IVIVE) leverage in vitro responses to predict the corresponding in vivo exposures and thresholds of concern. The integrated approach is also expected to offer the potential for efficient tools to provide estimates of chemical toxicity to various wildlife species instead of animal testing. However, developing fish physiologically based toxicokinetic (PBTK) models for IVIVE in ecological applications is challenging, especially for plausible estimation of an internal effective dose, such as fish equivalent concentration (FEC). Here, a fish PBTK model linked with the IVIVE approach was established, with parameter optimization of chemical unbound fraction, pH-dependent ionization and hepatic clearance, and integration of temperature effect and growth dilution. The fish PBTK-IVIVE approach provides not only a more precise estimation of tissue-specific concentrations but also a reasonable approximation of FEC targeting the estrogenic potency of endocrine-disrupting chemicals. Both predictions were compared with in vivo data and were accurate for most indissociable/dissociable chemicals. Furthermore, the model can help determine cross-species variability and sensitivity among the five fish species. Using the available IVIVE-derived FEC with target pathways is helpful to develop predicted no-effect concentration for chemicals with similar mode of action and support screening-level ecological risk assessment.

In vitro to in vivo extrapolation for predicting human equivalent dose of phenolic endocrine disrupting chemicals: PBTK model development, biological pathways, outcomes and performance

In vitro to in vivo (IVIVE) leverages in vitro high-throughput biological responses to predict the corresponding in vivo exposures and further estimate the human safe dose. However, for phenolic endocrine disrupting chemicals (EDCs) linked with complicated biological pathways and adverse outcomes (AO), such as bisphenol A (BPA) and 4-nonylphenol (4-NP), plausible estimation of human equivalent doses (HED) by IVIVE approaches considering various biological pathways and endpoints is still challenging. To explore the capabilities and limitations of IVIVE, this study conducted physiologically based toxicokinetic (PBTK)-IVIVE approaches to derive pathway-specific HEDs using BPA and 4-NP as examples. In vitro HEDs of BPA and 4-NP varied in different adverse outcomes, pathways, and testing endpoints and ranged from 0.0013 to 1.0986 mg/kg bw/day and 0.0551 to 1.7483 mg/kg bw/day, respectively. In vitro HEDs associated with reproductive AOs initiated by PPARα activation and ER agonism were the most sensitive. Model verification suggested the potential of using effective in vitro data to determine reasonable approximation of in vivo HEDs for the same AO (fold differences of most AOs ranged in 0.14-2.74 and better predictions for apical endpoints). Furthermore, system-specific parameters of cardiac output and its fraction, body weight, as well as chemical-specific parameters of partition coefficient and liver metabolic were most sensitive for the PBTK simulations. The results indicated that the application of fit for-purpose PBTK-IVIVE approach could provide credible pathway-specific HEDs and contribute to high throughput prioritization of chemicals in a more realistic scenario.