Deep Learning Decoding Algorithm

---

Code Structure

---

1. model_ResUNet ->Contains the neural network architecture code. ->No modification required.

2. utils ->Configuration and helper functions required for training. ->No modification required.

3. train_DecodeNet ->Training script for the neural network. ->Modify the following parameters within this script: Number of training epochs, Learning rate, Batch size, Output path for results.

4. test_DecodeNet ->Testing/evaluation script for the trained network. ->Modify the following parameters within this script: Path to the trained model weights, Input and output paths for testing data.

---

Training Data Preparation

---

Prepare input-output pairs as 4-dimensional npy files with dimensions: [N, C, W, H] N: Number of training samples C: Number of channels (e.g., C=4 for 4-color fluorescence images) W: Width of the mosaic (default: 512 pixels) H: Height of the mosaic (default: 512 pixels)

---

Dataset Configuration

---

1. my_dataloader ->Data loading script for training. ->Modify the following within this script: Name/path of the training dataset npy file, Normalization parameters.

---

Usage Notes

---

1. Ensure all file paths and parameters are correctly set before running train_DecodeNet or test_DecodeNet.

2. Default parameters are optimized for 512×512 pixel inputs. Adjust dimensions in model_ResUNet if using non-default sizes.