Termitomyces albuminosus is one species of agaric fungus in the family Agaricaceae in food production. However, it is a high demand to increase the current model detection accuracy, particularly under the complex planting environment, such as the variety of light and dark, the low recognition of soil and termitomyces albuminosus, the dense growth distribution and the serious shelter. In this study, target detection was proposed using an improved YOLOv5s. Firstly, RFBSE (Receptive Field Block Squeeze and Excitation) module was integrated into the backbone network. The human sensory field was then simulated to enhance the consistent contribution of different pixels to neural nodes in the process of network feature extraction. The edge features were highlighted to focus on the key areas. The irrelevant disturbances were suppressed using the module, such as the background. The attention mechanism of the channel was applied to gain the weight of different channels for the adaptive calibration of channel characteristic response. The channel was also enhanced to contain the important characteristic information of termitomyces albuminosus. As such, the high characteristic expression was achieved in the termitomyces albuminosus. Secondly, a multi-branch sampling DCSPP (Double Conv Spatial Pyramid Pooling) Pooling module was designed to perform the multiple sampling, in order to fuse the multiple receptive fields and then strengthen the relation between local and global information. The expression ability of the feature layer was enriched to improve the detection accuracy. Thirdly, the RFP (Recursive Feature Pyramid) structure was adopted in the neck network. The number of samples cannot increase, due to the usual interclass occlusion between termitomyces albuminosus. Previously, the network paid attention to the same image twice, because the context information around the occlusion samples was very important, and the feedback feature layer generated in the FPN structure was re-fed back to the backbone network for the Recursive computation. The neuronal activation was able to learn the correspondence and selectively inhibit, in order to improve the detection ability of the dense occluded sample of termitomyces albuminosus. The semantic information transmission was also promoted to enhance the context information near the occluded sample. At the same time, the cascaded RFP structure and the network fusion structure were lightened to reduce the calculation of parameters and memory usage. The ablation results showed that the RFBSE module, multi-branch pool module, and recursive pyramid structure shared different effects on the model. Specifically, the average precision mAP, precision, and recall rate of the final model reached 90.8%, 86.5%, and 84.8%, respectively. Each index of the improved model was improved, compared with the original one. The higher quality of the bounding box was achieved to detect the occluded target, where the mAP, accuracy, and recall were improved by 2.7, 3.8, and 3.9 percentage points, respectively. At last, the detection test of termitomyces albuminosus was carried out in different environments and shelter conditions by hardware platform model deployment. The visualized results showed that the detection rate of the model was more than 90%, which verified the validity of the model. The experimental results show that the improved model can be expected to accurately and rapidly identify the termitomyces albuminosus in a complex environment. The finding can provide technical support for the development of the termitomyces albuminosus harvesting robots.

Dragon fruit (Hylocereus spp.) is one of the most favorite tropical crops in the food processing industry worldwide, due to its vibrant appearance, sweet flavor, and nutritional profile rich in dietary fiber, vitamins, antioxidants, and natural pigments. The cultivation area has expanded to over 700,000 acres by 2021, thereby yielding more than 1.6 million tons annually in China. Yet the low utilization is only approximately 8% after processing. Separation techniques are often required for the product quality in juices, purees, jams, and wines. Conventional mechanical processing, such as roller extrusion or screw pressing, has frequently caused high pulp residue, peel fragmentation, pigment migration, and microbial contamination, leading to low juice yield, purity degradation, sensory defects, and elevated sanitation costs. Manual extraction can also induce high labor intensity, inconsistency, and hygiene risks. In this study, the high-pressure gas jet was introduced for the non-contact peel-pulp separation, in order to enhance the value-added applications, including the peel-derived red betacyanin pigments for the food colorants. Process parameters were also optimized to minimize the fruit damage for high efficiency. According to fluid mechanics, a jet impact force model was formulated to incorporate the velocity derived from pressure differentials via Bernoulli's equation, the impact force as the momentum flux proportional to the jet area and velocity squared, shear stress at the interface influenced by dynamic viscosity and velocity gradient, and exponential decay with the standoff distance to account for jet attenuation. Key factors were also identified, including the jet pressure (P_s), standoff distance (x), and nozzle diameter (d). The results revealed that the increasing P_s enhanced the momentum, the d induced the intensity bidirectionally, and the x induced the decay and diffusion. The normal stress and tangential shear were collectively modulated for the effective detachment without rupture. Single-factor experiments showed that there were the trends of the five levels: The jet pressure from 0.3 to 0.7 MPa exhibited the rising pulp removal rates peaking at 0.6 MPa with 90.31% efficiency, but declining sensory scores beyond, due to the excessive force inducing tears; The standoff distance from 10 to 50 mm shared the optimal balance at 30 mm, thus yielding 88.13% removal and 4.40 sensory score. The shorter distances risked the fragmentation, while the longer ones reduced the impact. The nozzle diameter from 1 to 5 mm was maximized at 4 mm with 88.57% removal and 4.36 score, in order to avoid the insufficient coverage at smaller sizes or diluted force at larger ones. A Box-Behnken response surface method facilitated the multifactor optimization. Three levels per parameter were employed to generate 17 experimental runs. Pulp removal rate (Y₁) and sensory score (Y₂, assessed via a 5-point hedonic scale for the pulp integrity, peel wholeness, residue, and color) were served as the responses. The quadratic polynomial regression was obtained with the high coefficients of determination (R²=0.979 3 for Y₁, 0.981 8 for Y₂). Analysis of variance highlighted that the most significant influencing factors on the model accuracy were ranked in descending order of the nozzle diameter, standoff distance, and jet pressure. Particularly, there was a great correlation between standoff distance and nozzle diameter, indicating the nonlinear separation. Multi-objective optimization determined the optimal conditions: the jet pressure 0.6 MPa, standoff distance 30 mm, nozzle diameter 4 mm, predicting 87.24% pulp removal and 4.509 sensory score. Verification trials were achieved in the 90.46% removal and 4.510 score, with the discrepancies under 5%, indicating the high prediction accuracy. Incorporating dragon fruit's material properties—high moisture content, viscoelastic pulp, thin elastic peel, and curved geometry—elucidated mechanisms: gas jets induce mixed-mode fracture (Mode I opening via normal impact, Mode II shear along interface) to overcome pectin-mediated adhesion. This optimized non-contact process not only elevates pulp recovery and sensory attributes but also supports automated equipment development for flexible fruits and vegetables, fostering industrial scalability, reduced waste, and full-value utilization in sustainable food systems.

Channa argus is one type of predatory fish in the family Channidae native to fresh waters in eastern China. The skin of Channa argus has been an ideal substitute for the shortage of wild high-grade leather raw materials, due to the beautiful pattern, strong toughness, excellent breathability, waterproof, and strong anti-expansion. Annual production of Channa argus skin has been more than 500 000 tons in the broad market of China. However, the manual peeling of Channa argus skin has been limited to the deep processing industry in large-scale production at present, leading to low efficiency, low integrity of fish skin, and security risks. Therefore, this study aims to design the automatic whole-skin peeling system of Channa argus using the 3D (Three-Dimensional) reconstruction. The morphological parameters were also collected, where the head, body and tail sections were changed dramatically. The sticky skin was then fixed for the peeling operation on the hard hip fin. The whole-skin peeling machine was designed, including clamping, cutting, skin and flesh separation, as well as peeling devices. The vision system was selected with the industrial camera, LED (light-emitting diode) lamp band and upper computer. The industrial camera was used to capture the images of Channa argus. The LED lamp band was to provide an ideal optical environment for image acquisition. The upper computer was also used to preprocess the images. 3D reconstruction of images was performed to identify and locate the pectoral fin position using YOLOv5s. The cloacal aperture position was then identified and located by ROI (region of interest). The operation path of each mechanism was finally realized, according to the 3D point cloud model. The control system was designed to accurately control the travel mechanisms and the actuators in the whole-skin peeling device. The position feedback PID (Proportion Integral Derivative) with magnetic encoder was used for the closed-loop control of the motion module, in order to avoid the out-of-step phenomenon during operation. In the actuators, the air path was designed to realize the precise control of the cylinder gripper, horizontal slide cylinder, longitudinal slide cylinder, and double rod double shaft cylinder, according to the control requirements of pneumatic components. The duty ratio of the PWM (pulse width modulation) signal was set to realize the control of a DC (direct current) motor speed, in order to calibrate the relationship between PWM duty ratio and rotational speed. The performance of the prototype was evaluated by the total system operation duration, fish skin removal rate, skin and flesh separation index and breakage rate of fish skin of the whole-skin peeling device. The test results showed that a high peeling speed was achieved in the machine. The total operation duration of the system varied with the size of the Channa argus, ranging from 172 to 195 s, with an average of 183 s. After peeling, the fish skin removal rate ranged from 93.10% to 95.70%, with an average value of 94.60%. There was a large actual area of Channa argus skin after stripping, which effectively reduced the waste of raw materials. The separation index of skin and meat ranged from 9.91 to 9.98, with an average value of 9.93. There was less fish left on the skin after stripping, indicating a higher stripping efficiency than before. The breakage rate of fish skin ranged from 5.00% to 10.00%, with an average of only 6.25%. The small number of damaged fish skin in each group fully met the requirements of whole-skin peeling technology. This finding can provide a strong reference for the development and design of the whole-skin peeling system of Channa argus.

A fairly high breakage rate can often occur in the cold-chain logistics of fresh eggs. In this study, the compression and freshness detection were designed for fresh eggs, with the storage temperature of the eggs and the coating material on the surface as the main parameters, and the storage time as the boundary conditions. The results demonstrated that the temperature of 2-8 ℃ retarded the rottenness of eggs for better pressure resistance. Whether at room temperature or low temperature, the relative elastic deformation with time was maintained at a high value. There was a small variation in the type of coating agent. The rottenness of eggs after coating was decelerated at room temperature. While the pressure resistance of eggs was improved to compromise at 2-8 ℃. Both the coating agent and the low temperature were superimposed to slow down the decrease of its freshness. A professional experimental group was organized to conduct the sensory scoring on different groups of eggs under various storage times. The Huff unit value and sensory scoring were similar to each other. Therefore, the freshness of fresh eggs decreased with the extension of storage time. After 14 days of storage, a small amount of egg contents were solidified in the non-coated group of eggs stored at 2 ℃. The sensory scores of the group were lower during that time period. A comprehensive evaluation model was established for the parameters of the cold-chain transportation using the analytic hierarchy process, particularly with the ultimate load, the relative elastic deformation, the Haugh unit value, and the sensory score of eggs as the decision-making indicators. The relative importance of each index was ranked to determine the importance discrimination matrix. The weights were calculated corresponding to each decision index, thus reflecting the importance of the decision index to the final. Four decision indicators were ranked in descending order of the Hough unit value, sensory score, ultimate load, and relative elastic deformation, according to the priority. The optimal parameters were achieved in the cold-chain logistics temperature and the coating material for different transportation durations as follows. In the cold-chain logistics duration of 0-4 days, the transportation temperature was at 2 ℃ with no paints applied; In the duration of 4-7 days, the transportation temperature was 5 ℃, with the chitosan solution chosen as the paints. In the duration of 7-10 days, the transportation temperature was selected at 2℃ with the polyvinyl alcohol solution chosen as the paint. For the duration of 10 to 14 days, the transportation temperature was 5 ℃ with no paints applied. The quality of stored eggs was generally better at low temperatures than that at room temperature. The coated eggs at room temperature were generally better than the uncoated eggs. While the coated eggs at low temperature were less than the uncoated eggs. The coating layer protected the eggs to reduce the damage from the external physical collisions in the cold-chain logistics of a short or medium distance, indicating the less rottenness of eggs. However, the coating layer on the eggs failed to protect them in the process of long-term cold-chain transportation. The coating layer deteriorated and even fell off in the low-temperature environment for a long time, leading to less pressure resistance and maintenance of the egg quality. The findings can provide theoretical support for the cold-chain logistics of fresh eggs.

Aiming at the problems of traditional manual grading of stropharia rugoso-annulata, such as high labor intensity, low efficiency, and poor consistency, an improved method based on MobileViT model was proposed. By designing multi-scale modules with adaptive branching, adding local and global feature fusion, and introducing dual attention modules, the feature extraction capability is improved and the model robustness is enhanced. The experimental results show that the average recognition accuracy of the improved XCA-MobileViT for the five levels of stropharia rugoso-annulata datasets on the experimental platform is 97.71%, which is 2.34% higher than that of the MobileViT model, and the number of parameters and computation decreased by 0.401 M and 140.2 M respectively. Through validation experiments on two publicly available datasets of mushrooms, it was found that the F1 score and accuracy of XCA-MobileViT exceeded other models compared and showed good generalization.