train multiple channel data using YOLO

Refer to:

https://blog.csdn.net/Lizongming_/article/details/135652383

1. Run python file to train model

from ultralytics import YOLO

# Create a new YOLO model from scratch
#model = YOLO('yolov8n.yaml')
model = YOLO('yolov8x.yaml')

# Load a pretrained YOLO model (recommended for training)
model = YOLO('yolov8n.pt')

#model = YOLO('yolov8n.yaml').load('yolov8n.pt')
model = YOLO('yolov8x.yaml').load('yolov8n.pt')

# Train the model using the 'tree.yaml' dataset for 500 epochs
results = model.train(data='./Datasets/tree.yaml ', 
	epochs=500,
	iou=0.5,
	imgsz=64, 
	batch=2,
	lr0=0.001, 
	lrf=0.001,
	hsv_h=0,  #.015  # (float) image HSV-Hue augmentation (fraction),
	hsv_s=0,  #.7  # (float) image HSV-Saturation augmentation (fraction),
	hsv_v=0, #.4  # (float) image HSV-Value augmentation (fraction)
	workers=1,
	project='./word_dir/std_7bands_yolov8x',
	device=0)

# Evaluate the model's performance on the validation set
#results = model.val()

# Perform object detection on an image using the model
#results = model('https://ultralytics.com/images/bus.jpg')

# Export the model to ONNX format
#success = model.export(format='onnx')

把hsv_h，hsv_s和hsv_v三个参数设置为0，不做HSV数据增

2.  .yaml  to load data

path: /home/ultralytics/Datasets/7bands/  # dataset root dir
train: images/train/
val: images/val/
#test: ./Datasets/Tree/images/val/

# number of classes
nc: 1
#chanel number of the data
ch: 7

# class names
#names: [["pine", "spruce", "birch", "populus"]]
names: ["tree"] 

3. /ultralytics/data/utils.py 

def verify_image_label(args):

4. /ultralytics/data/base.py  

def load_image(self, i, rect_mode=True):

def affine_transform(self, img, border):

5. /ultralytics/data/augment.py

class LetterBox: def __call__(self, labels=None, image=None):

6. /ultralytics/utils/plotting.py 

def plot_images

class Annotator:

7. /ultralytics/cfg/default.yaml

channel: 7

8.

(1) in ./ultralytics/engine/predictor.py

def stream_inference(self, source=None, model=None, *args, **kwargs):

# Warmup model
            # Warmup model
            if not self.done_warmup:
                img_ch = 5
                self.model.warmup(imgsz=(1 if self.model.pt or self.model.triton else self.dataset.bs, img_ch, *self.imgsz)) #self.args.channel 3
                self.done_warmup = True

(2)  './ultralytics/data/build.py'

load_inference_source

(3) Predict: in Line 490,  './ultralytics/data/loaders.py'

class LoadImagesAndVideos,  def __next__(self):

            else:
                self.mode = "image"
                if path.endswith('.tif') or path.endswith('.TIF'):
                    im0 = tifffile.imread(path)
                else:
                    im0 = cv2.imread(path)  # BGR
                #im0 = cv2.imread(path)  # BGR
                if im0 is None:
                    raise FileNotFoundError(f"Image Not Found {path}")

save_pred

        # Save imgs
        if self.dataset.mode == "image":
            if im0.shape[2] >3:
                im0 = im0[:, :, :3]
            cv2.imwrite(save_path, im0)