Inputs for ANN (multihidden-layer feed-forward artificial neural network) models were drying time (t(i + 1)), initial temperature (T0), moisture content (MC0), microwave power, and vacuum pressure. The outputs were temperature (T(i + 1)) and moisture content (MC(i + 1)) at a given t(i + 1). After training the ANN models with experimental data using the Levenberg-Marquardt algorithm, a two-hidden-layer model (25-25) was determined to be the most appropriate model. The mean relative error (MRE) and mean absolute error (MAE) of this model for T(i + 1) were 1.53% and 0.77 degrees C, respectively. In the case of MC(i + 1), the MRE and MAE were 11.48% and 0.04 kg(water)/kg(dry), respectively. Using temperature (T(i)) and moisture content (MC(i)) values at t(i) in the input layer significantly reduced the computation errors such that MRE and MAE for T(i + 1) were 0.35% and 0.18 degrees C, respectively. In contrast, these error values for MC(i + 1) were 1.78% (MRE) and 0.01 kg(water)/kg(dry) (MAE). These results indicate that ANN models were able to recognize relationships between process parameters and product conditions. The model may provide information regarding microwave power and vacuum pressure to prevent thermal damage and improve drying efficiencies.

The image analysis technique and artificial neural networks (ANNs) for boiled shrimp's shape classification were developed in this research. A color image of boiled shrimp in red‐green‐blue format was processed and analyzed to determine the shape feature as a relative internal distance (RID). The RID was the ratio between the shortest distance measured perpendicularly between the center line and the shrimp's contour. The RID values from different 62 locations were calculated. The multilayer ANN models were trained to classify shapes of the boiled shrimp using the RID values as the network input. The analysis showed that the 15‐node ANN model was highly effective for boiled shrimp's shape classification with 99.80% overall accuracy (regular‐shaped shrimps [100.00%], shrimps with no tails [100.00%], with one tail [97.78%] and with broken body [100.00%]). The RID values were considered as an appropriate shape representation for boiled shrimps. The ANN model recognized most boiled shrimp's shape through the RID profiles. Practical Application The developed image analysis technique and artificial neural network model effectively classified the boiled shrimp's shape using the relative internal distance values. This technique can be used in the development of an automatic sorting system for boiled shrimp based on real‐time image processing.

The drying rate of a mushroom undergoing microwave-vacuum (MV) drying (MVD) was controlled by moisture dissipation and was dependent on vacuum pressure levels. The main objective of this work was to develop artificial neural network (ANN) model to predict moisture ratio of MV-dried mushrooms. One-hidden-layer feed-forward ANN models were trained and validated with experimental data. The Levenberg-Marquardt algorithm was utilized in regulating the ANN model weights and biases. Inputs for ANN models were vacuum pressure and drying time. Output from ANN models was moisture ratio at a given drying time. Reduced chi-square (X 2) and root mean square error (RMSE), and residual sum of squares (RSS) of the results from ANN models were calculated and compared with those of a modified Page's model (an experimental-based mathematical model), which is commonly used in the literature. The X 2, RMSE, and RSS of the ANN model (2.272 x 10 -5, 4.023 x 10 -3, and 3.204 x 10 -3, respectively) were found to be lower than those of the modified Page's model (6.692 x 10 -4, 2.561 x 10 -2, and 12.98 x 10 -2, respectively). These results indicate that the feed-forward ANN model represented the drying characteristics of mushrooms better than the modified Page's model. Therefore, the ANN model could be considered as a better tool for estimation of the moisture content of mushrooms than by the modified Page's model.

Temperature (T) and moisture content (MC) of non-homogenous food undergoing microwave-vacuum (MV) drying (MVD) are directly dependent on microwave power, vacuum pressure, and the product's physical properties. A two-hidden-layer Artificial Neural Network (ANN) model was developed in an earlier study to predict temperature and moisture content of the product at a given time based on the present state of product conditions and process control parameters. This approach either provided lowest error in temperature prediction or in moisture content prediction but not the lowest error in both the prediction parameters simultaneously. The main objective of this work was to improve the performance of the ANN model for temperature and moisture content predictions in MV dried samples. Experimental data obtained from MVD of tomato slices at different drying conditions was normalized and divided into two groups for training and validating. The parallel dynamic ANN model consisted of two double-hidden-layer feed-forward ANN models with varying node numbers (10, 20, and 30). These models were separately trained, simultaneously for moisture content as well as temperature, with the Levenberg-Marquardt algorithm. Inputs for the ANN models were magnetron on-off status, vacuum pressure, temperature, and moisture content at time`t i '. The previous temperature and moisture content data at time`t i-1, i-2, …, i-n ' where n = 0, 10, 20, and 30 were also added to the input layer. Outputs from the ANN models were temperature and moisture content at time`t i+1 '. The results indicated that the dynamic ANN model working in parallel with the previous temperature and moisture content data provided results that are more accurate and required less training time than those of ordinary ANN models. Model simulation may supply essential information regarding temperature and moisture content of non-homogenous foods corresponding to microwave power and vacuum pressure levels to the predictive control system. Therefore, improved drying efficiencies and thermal damage prevention may be achieved.

In an actual sorting process, shrimps are fed to a machine-vision-based sorter at random postures. This study proposed an Enhanced Artificial Neural Network (E-ANN) coupled with a Pattern Recognition ANN (P-ANN) model to overcome the posturespecificity of the regression ANN model commonly used for mass estimation of the headless-shell-on (HSO) shrimps. Images of 103 shrimps with seven different postures were used. The similarity of any shrimp image to the reference shrimp postures (i.e., extended-legs, collapsed-legs, curl body, and dorsal body) was determined by the P-ANN model and used as additional input, besides the area and perimeter. The coupled-ANN model could accurately estimate the mass of shrimps with random postures (R 2 = 0.70 to 0.88 and MRE = À2.62 to 2.97%) within~10 ms per shrimp, which is practical to use in an automatic shrimp sorting system based on machine vision technique. Further enhancement of the model performance could be achieved by adding color and texture features to distinguish different shrimp parts (e.g., body, legs, and tail). Practical ApplicationsThis research shows the potential of using a pattern recognition artificial neural network (P-ANN) to cut the limitation of the conventional regression ANN model based on area and perimeter for shrimp mass estimation, which is specific to shrimp posture (or shape). Since the coupled P-ANN and enhanced artificial neural network (E-ANN) used a relatively short processing time (~10 ms per shrimp) and yielded less than 1 g error in the mass estimation of the headless-shell-on (HSO) shrimps with random postures, it is possible to incorporate this coupled ANN model in an automatic shrimp sorting system based on machine vision technique. | INTRODUCTIONThe global farmed shrimp market continues to grow faster than those for other aquaculture species. The global trade of shrimp and prawns is estimated at USD 28 billion per year, coming mostly from farms in Asia and Latin America (mainly Ecuador). The majority of farmed shrimps is white shrimp or Penaeus vannamei (FAO, 2020). Since fresh shrimps are easily perishable, they are frozen and shipped from countries of origin to customers around the world. The shrimp products are available in various forms such as head-on-shell-on (HOSO) shrimp, headless-shell-on (HSO) shrimp, deveined shrimp, and so on. The frozen HSO shrimp is one of the highly sought-after export products of Thailand. Production of frozen HSO shrimps starts with receiving raw fresh shrimps from suppliers. The shrimps are then cleaned and de-headed. The removal of shrimp heads is a crucial step to prolong their shelf-life because shrimp heads contain several