Supplementary data and algorithms associated with the article of deep learning for predicting TVC in peeled shrimp

Python codes:
1.SAEs.py #this file is used to train the SAEs model
2.SAEs-PLSR.py #use this file to train PLSR model based on deep spectra features, and evaluate the model.
3.SPA-PLSR.py #use this file to train PLSR model based on characteristic wavelengths selected by SPA, and evaluate the model.
4.F-PLSR.py #use this file to train PLSR model based on full spectra, and evaluate the model.
Data:
1.ramdonpixel_1.pkl #60060 spectra for training SAEs model
2.ramdonpixel_2.pkl #60060 spectra for validating SAEs model
3.Fullspectra.csv #200 samples with 230 bands spectra and reference TVC values
4.SPAspectra.csv #200 samples with 18 characteristic wavelengths and reference TVC values
Logs:
saved model files
Results:
experimental results files

Python代码：
1. SAEs.py（此文件用于训练自编码器模型）
2. SAEs-PLSR.py（使用此文件基于深度光谱特征训练偏最小二乘回归模型，并评估模型）
3. SPA-PLSR.py（使用此文件基于通过SPA选择的特征波长训练偏最小二乘回归模型，并评估模型）
4. F-PLSR.py（使用此文件基于完整光谱训练偏最小二乘回归模型，并评估模型）
数据：
1. ramdonpixel_1.pkl（用于训练自编码器模型的60060条光谱）
2. ramdonpixel_2.pkl（用于验证自编码器模型的60060条光谱）
3. Fullspectra.csv（200个样本，包含230个波段的光谱和参考TVC值）
4. SPAspectra.csv（200个样本，包含18个特征波长和参考TVC值）
日志：
saved model files（保存的模型文件）
结果：
experimental results files（实验结果文件）