To address the insufficient acquisition of key growth data in Litopenaeus vannamei aquaculture, we designed an adjustable intelligent detection system based on a feeding tray for the acquisition and identification of residual feed and shrimp body length information. The system used a mobile cart as a platform, adjusted the horizontal distance of the feed tray through a retractable cantilever, and employed an ultrasonic sensor combined with a PID control algorithm to achieve precise hovering of the feed tray. The device was equipped with a vision sensor to capture images of the feed tray area, utilizing image processing algorithms to automatically detect shrimp and residual feed. The results indicate that the success rates of hovering at water depths of 0.50 m (±0.01 m) and 0.05 m (±0.01 m) reached 96.7% and 93.3%, respectively; while the relative errors of shrimp body length measurement and residual feed counting were both no more than 6%. This reliable system with precise visual detection offers a viable solution for reducing manual pond inspection intensity and enhancing mechanized and precision feeding in aquaculture.

Feeding equipment has been restricted to the feeding ranges and pond transfers, such as the fixed shore-based feeders and feeding boats. This study aims to optimize the materials used in the feed-dispensing mechanism for the pneumatic feeders. The mobile carts were also utilized to achieve the flexible pond transfers and uniform feeding around the pond perimeter. An adjustable feeding mechanism was incorporated for the precise feed dispensing within the required target range. The feed dispensing of pneumatic vehicles significantly improved the feeding efficiency and feed distribution, compared with the existing feeding devices in practical applications. The coupled EDEM-FLUENT simulation was carried out to optimize the performance of the feeding mechanism. The key parameters were determined on the feed dispersion. Specifically, four parameters were selected: the fan wind speed, the angle of the feeding pipe, the length of the feeding pipe, and the diameter of the feeding pipe. A series of single-factor experiments was conducted to determine the impact of each parameter on the feeding performance under different conditions, thus providing a basis for further optimization. An orthogonal experiment was also conducted using the data from these single-factor experiments. The optimal combination of the key parameters was obtained using the response surface method (RSM). The results showed that the better feeding performance of the pneumatic feed dispensing vehicle was achieved in the optimal feed coverage and uniform distribution within a 3-5 m range, when the fan wind speed was 16.1 m/s, the feeding pipe's angle was 9°, the feeding pipe length was 425 mm, and the diameter was 42 mm. In actual prototype testing, the dispersion of the pneumatic feed dispensing vehicle was very close to the simulation, with the relative errors in the coverage and uniformity within the feeding range of 5.4% and 6.5%, respectively. These small error ranges demonstrated that the pneumatic feed dispensing vehicle effectively met the requirements of precise and uniform feed dispersion within a set range. The simulation model was verified for the feasibility of the experiment. The flexible movement of the pneumatic feed dispensing vehicle was found within shrimp ponds after optimizations, in order to precisely control the feed dispersal range and distribution. There was uniform coverage within the 3-5 m range. The feed vehicle was adaptable to different requirements of the shrimp pond, especially in the large aquaculture farms that required the larger feeding areas. A more efficient solution was provided after verification, compared with the conventional equipment. Moreover, the pneumatic feed dispensing vehicle was also featured by the multi-pond functionality, thus meeting the feeding needs of multiple ponds. The vehicle's high flexibility and adjustability can avoid purchasing separate feeding devices for each pond, thereby saving significant equipment investment and maintenance costs. At the same time, the key parameters were adjusted, such as the fan wind speed, feeding pipe length, diameter, and inclination angle. The pneumatic feed dispensing vehicle can also be customized for optimal feeding performance, according to different farming environments and needs.

Shrimp has been one of the most favorite seafood for years. It is essential to timely estimate the amount of feed left in the bait tray after feeding in shrimp aquaculture. Feeding strategies can also be adjusted to reduce the bait costs in recent years. The traditional detection of residual baits can rely on visual inspection by the shrimp farmers. Neural networks and deep learning have been introduced to detect and count the residual baits at present. However, large-scale neural networks cannot be successfully implemented on mobile devices, due mainly to the low recognition accuracy and large model computation. In this study, the improved model was proposed to estimate the density map of residual baits using a hybrid dilated convolution and attention multi-scale network (HAMNet). The high accuracy and low complexity were achieved in the detection models of residual bait. The HAMNet model was divided into three components: A low-level feature extractor (LLFE), a high-level feature extractor (HLFE), and a density map restorer and generator (DMRG). These components served as the front-end, the middle-end, and the back-end network of the improved model, respectively. Firstly, inspired by the multi-column convolutional neural network (MCNN), the parallel convolution block (PCB) was designed in the front-end network, in order to extract the feature information of residual bait at multiple scales within a single-column architecture; At the same time, the hybrid dilated convolution block (HDCB) was introduced into the mid-end network to expand the receptive field, in order to further learn the multi-scale features. Secondly, a channel attention mechanism (CAM) was embedded into the network to recalibrate the weights of useful feature information using the interdependence among channels, in order to highlight the difference between the target and background. Finally, the learnable transposed convolutional layers were applied in the back-end network to recover the detailed information from the feature maps. The quality of density maps was improved to reduce the counting errors. As such, the high-quality density map was then obtained during downsampling in the front-end network. In addition, the effectiveness of the improved model was validated using residual bait images under bait tray conditions. A comparison was also implemented with the classical networks of density map estimation. Comparative experiments showed that the HAMNet model was achieved in the minimum mean absolute error (MAE) of 2.0, the minimum root mean square error (RMSE) of 2.9, and the least floating point operations (FLOPs) at 6.55 G on the residual bait datasets, with a parameter count of only 0.52MB. The HAMNet model shared the higher counting accuracy and stability with the lower computational complexity. Compared with the baseline MCNN network, the improved model achieved a 44.4% reduction in MAE, a 40.8% reduction in RMSE, and a 13.7% reduction in FLOPs. Compared with the CMTL, SANet, and CSRNet, the optimal balance was obtained in all performance metrics. In summary, the HAMNet model outperformed the rest, in terms of overall performance, thus improving the counting accuracy with the low computational volume. The finding can provide a strong reference to rapidly quantify the residual baits in shrimp aquaculture. Novel ideas were also offered to deploy the detection models of residual bait on the platforms with limited computational power.