In the production process from green beans to coffee bean packages, the defective bean removal (or in short, defect removal) is one of most labor-consuming stages, and many companies investigate the automation of this stage for minimizing human efforts. In this paper, we propose a deep-learning-based defective bean inspection scheme (DL-DBIS), together with a GAN (generative-adversarial network)-structured automated labeled data augmentation method (GALDAM) for enhancing the proposed scheme, so that the automation degree of bean removal with robotic arms can be further improved for coffee industries. The proposed scheme is aimed at providing an effective model to a deep-learning-based object detection module for accurately identifying defects among dense beans. The proposed GALDAM can be used to greatly reduce labor costs, since the data labeling is the most labor-intensive work in this sort of solutions. Our proposed scheme brings two main impacts to intelligent agriculture. First, our proposed scheme is can be easily adopted by industries as human effort in labeling coffee beans are minimized. The users can easily customize their own defective bean model without spending a great amount of time on labeling small and dense objects. Second, our scheme can inspect all classes of defective beans categorized by the SCAA (Specialty Coffee Association of America) at the same time and can be easily extended if more classes of defective beans are added. These two advantages increase the degree of automation in the coffee industry. The prototype of the proposed scheme was developed for studying integrated tests. Testing results of a case study reveal that the proposed scheme can efficiently and effectively generate models for identifying defective beans with accuracy and precision values up to $80\%$ .

Single-image haze removal algorithms restore clear images from hazy inputs but often struggle with high computational complexity and halos or artifacts from patch-based priors. We propose a novel non-local patch-based haze removal algorithm that estimates atmospheric light and scene transmission using the minimum image channel. This approach reduces reliance on patch-based priors, ensuring efficient and high-quality dehazing. An adaptive gain, based on the ratio of Imin and Imax, adjusts the minimum channel to refine atmospheric light, scene depth, and transmission estimates, addressing inaccuracies in traditional methods. Histogram equalization and image multiplication enhance scene radiance, improving contrast, color accuracy, and naturalness under challenging haze conditions. Subjective evaluations confirm our method achieves the highest MOS, consistently rated most visually appealing. In addition, objective metrics show our method outperforms state-of-the-art techniques, achieving better FADE, NIQE, SSEQ, BRISQUE, and PIQE scores by 2% to 13%, indicating more natural colors, minimal distortion, and superior restoration quality. Additionally, our method demonstrates the lowest computational cost and processes 8K images up to 21 times faster than the second-fastest method, CAP. This breakthrough makes the algorithm ideal for real-time applications in consumer electronics such as smartphones and digital cameras, delivering high-quality, haze-free imagery with minimal computational cost.

Surgical smoke generated by energy devices poses health risks to medical staff. During laparoscopic surgery, the smoke aggregating around the camera obstructs the visual field, forcing surgeons to interrupt surgery, and may increase surgical risk. We propose a proximal smoke evacuation method to improve surgical quality by effectively eliminating surgical smoke. A smoke evacuation device was designed to attach to a laparoscopic electrode and create a channel for smoke to be suctioned directly from near the tip of the electrode (proximal evacuation). An animal study was conducted to collect videos of electrocautery with proximal (device) evacuation, distal (trocar) evacuation, and no evacuation. We used a computer vision-based model to compare in-screen smoke density and image quality between different evacuation pathways. Compared with distal and no evacuation, proximal evacuation had significantly lower estimated in-screen smoke density and higher image quality (p < 0.001). The pneumoperitoneum pressure remained above 8 mmHg throughout the procedure with proper suction pressure setting. A total of 62 trials performed by 15 surgeons produced consistent results, supporting the core findings of this study. On average, proximal evacuation can eliminate 85.47% of the smoke around operative fields within 10 s. Proximal smoke evacuation is capable of maintaining a clean surgical field and high image quality during laparoscopic surgery. The device can help avoid interrupting surgeries to wait for the smoke to clear or clean the camera lens.

This paper presents an artificial intelligence (AI)-based method for surface temperature prediction, applying a hybrid convolutional neural network and Lasso regression (CNN-Lasso) to thermocouple-calibrated thermal infrared (IR) images of a large cylindrical top-dome hollow porous radiant burner for the first time. Three CNN architectures, VGG-16, ResNet-50, and DenseNet-121, are evaluated as feature extractors within this framework. Applying the lean premixed CH4/air mixture at a fixed equivalent ratio  = 0.65, the burner is operated at an internal combustion mode over a range of power levels varying from 5 kW to 12 kW controlled by various flow rates of the pre-mixture. The corresponding surface temperature distributions of the porous radiant burner are recorded by an IR camcorder, calibrated by three K-type thermocouples positioned at the top-dome, middle, and bottom of the cylindrical hollow porous burner. The temperature measurement range and the emissivity of the IR camcorder are set at 300-2000°C and 0.95, respectively, having an average temperature of about 40°C difference/uncertainty as compared to that measured by the thermocouples. We evaluate the three CNN-Lasso hybrid models using 900 IR images with 9-fold cross-validation for training and the remaining 100 images for testing. All models exhibit stable and high prediction performance, achieving R² values above 0.85 on the test set, thereby confirming the effectiveness and robustness of the proposed framework. Thus, the proposed AI-driven framework introduces an innovative and accurate approach to non-intrusive surface temperature measurements using infrared imaging.