Description
I followed the official tutorial. 1. Train coco128. I started training map=0 from 0. 2. If I load the pre-trained weights of yolov13, the map would be 0.7, but I don't know if the network is trained. 3. Then I used my own dataset to load the yolov13 pre-trained weights, with a map of 0.000 or so, but it couldn't be improved. Why is this?
PS C:\Users\29363\Desktop\yolov13-main> & C:/Users/29363/AppData/Local/conda/conda/envs/yolov13/python.exe c:/Users/29363/Desktop/yolov13-main/train.py
New https://pypi.org/project/ultralytics/8.3.159 available 😃 Update with 'pip install -U ultralytics'
Ultralytics 8.3.63 🚀 Python-3.11.13 torch-2.7.1+cu128 CUDA:0 (NVIDIA GeForce RTX 5070 Laptop GPU, 8151MiB)
engine\trainer: task=detect, mode=train, model=yolov13s.yaml, data=coco128.yaml, epochs=20, time=None, patience=100, batch=8, imgsz=640, save=True, save_period=-1, cache=False, device=None, workers=4, project=None, name=train16, exist_ok=False, pretrained=True, optimizer=auto, verbose=True, seed=0, deterministic=True, single_cls=False, rect=False, cos_lr=False, close_mosaic=10, resume=False, amp=True, fraction=1.0, profile=False, freeze=None, multi_scale=False, overlap_mask=True, mask_ratio=4, dropout=0.0, val=True, split=val, save_json=False, save_hybrid=False, conf=None, iou=0.7, max_det=300, half=False, dnn=False, plots=True, source=None, vid_stride=1, stream_buffer=False, visualize=False, augment=False, agnostic_nms=False, classes=None, retina_masks=False, embed=None, show=False, save_frames=False, save_txt=False, save_conf=False, save_crop=False, show_labels=True, show_conf=True, show_boxes=True, line_width=None, format=torchscript, keras=False, optimize=False, int8=False, dynamic=False, simplify=True, opset=None, workspace=None, nms=False, lr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.0, box=7.5, cls=0.5, dfl=1.5, pose=12.0, kobj=1.0, nbs=64, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=0.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, bgr=0.0, mosaic=1.0, mixup=0.0, copy_paste=0.1, copy_paste_mode=flip, auto_augment=randaugment, erasing=0.4, crop_fraction=1.0, cfg=None, tracker=botsort.yaml, save_dir=runs\detect\train16
from n params module arguments
0 -1 1 928 ultralytics.nn.modules.conv.Conv [3, 32, 3, 2]
1 -1 1 9344 ultralytics.nn.modules.conv.Conv [32, 64, 3, 2, 1, 2]
2 -1 1 20800 ultralytics.nn.modules.block.DSC3k2 [64, 128, 1, False, 0.25]
3 -1 1 37120 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2, 1, 4]
4 -1 1 78464 ultralytics.nn.modules.block.DSC3k2 [128, 256, 1, False, 0.25]
5 -1 1 68352 ultralytics.nn.modules.conv.DSConv [256, 256, 3, 2]
6 -1 2 677120 ultralytics.nn.modules.block.A2C2f [256, 256, 2, True, 4]
7 -1 1 134400 ultralytics.nn.modules.conv.DSConv [256, 512, 3, 2]
8 -1 2 2664960 ultralytics.nn.modules.block.A2C2f [512, 512, 2, True, 1]
9 [4, 6, 8] 1 1343744 ultralytics.nn.modules.block.HyperACE [256, 256, 1, 8, True, True, 0.5, 1, 'both']
10 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
11 9 1 132096 ultralytics.nn.modules.block.DownsampleConv [256]
12 [6, 9] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
13 [4, 10] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
14 [8, 11] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 12] 1 0 ultralytics.nn.modules.conv.Concat [1]
17 -1 1 443648 ultralytics.nn.modules.block.DSC3k2 [768, 256, 1, True]
18 [-1, 9] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
19 17 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
20 [-1, 13] 1 0 ultralytics.nn.modules.conv.Concat [1]
21 -1 1 131712 ultralytics.nn.modules.block.DSC3k2 [512, 128, 1, True]
22 10 1 33024 ultralytics.nn.modules.conv.Conv [256, 128, 1, 1]
23 [21, 22] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
24 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
25 [-1, 18] 1 0 ultralytics.nn.modules.conv.Concat [1]
26 -1 1 345344 ultralytics.nn.modules.block.DSC3k2 [384, 256, 1, True]
27 [-1, 9] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
28 26 1 590336 ultralytics.nn.modules.conv.Conv [256, 256, 3, 2]
29 [-1, 14] 1 0 ultralytics.nn.modules.conv.Concat [1]
30 -1 1 1346048 ultralytics.nn.modules.block.DSC3k2 [768, 512, 1, True]
31 [-1, 11] 1 1 ultralytics.nn.modules.block.FullPAD_Tunnel []
32 [23, 27, 31] 1 850368 ultralytics.nn.modules.head.Detect [80, [128, 256, 512]]
YOLOv13s summary: 648 layers, 9,055,527 parameters, 9,055,511 gradients, 21.2 GFLOPs
Freezing layer 'model.32.dfl.conv.weight'
AMP: running Automatic Mixed Precision (AMP) checks...
AMP: checks passed ✅
train: Scanning C:\Users\29363\Desktop\yolov13-main\datasets\coco128\labels\train2017.cache... 126 images, 2 backgrounds, 0 corrupt: 100%|██████████| 128/128 [00:00<?, ?it/s]
albumentations: Blur(p=0.01, blur_limit=(3, 7)), MedianBlur(p=0.01, blur_limit=(3, 7)), ToGray(p=0.01, method='weighted_average', num_output_channels=3), CLAHE(p=0.01, clip_limit=(1.0, 4.0), tile_grid_size=(8, 8)), ImageCompression(p=0.5, compression_type='jpeg', quality_range=(75, 100))
val: Scanning C:\Users\29363\Desktop\yolov13-main\datasets\coco128\labels\train2017.cache... 126 images, 2 backgrounds, 0 corrupt: 100%|██████████| 128/128 [00:00<?, ?it/s]
C:\Users\29363\AppData\Local\conda\conda\envs\yolov13\Lib\site-packages\albumentations\check_version.py:147: UserWarning: Error fetching version info The read operation timed out
data = fetch_version_info()
Plotting labels to runs\detect\train16\labels.jpg...
optimizer: 'optimizer=auto' found, ignoring 'lr0=0.01' and 'momentum=0.937' and determining best 'optimizer', 'lr0' and 'momentum' automatically...
optimizer: AdamW(lr=0.000119, momentum=0.9) with parameter groups 139 weight(decay=0.0), 189 weight(decay=0.0005), 153 bias(decay=0.0)
Image sizes 640 train, 640 val
Using 4 dataloader workers
Logging results to runs\detect\train16
Starting training for 20 epochs...
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
1/20 3.34G 3.513 5.717 4.266 93 640: 100%|██████████| 16/16 [00:20<00:00, 1.29s/it]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:03<00:00, 2.11it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
2/20 3.48G 3.617 5.71 4.264 132 640: 100%|██████████| 16/16 [00:07<00:00, 2.26it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 2.99it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
3/20 3.39G 3.505 5.699 4.24 157 640: 100%|██████████| 16/16 [00:07<00:00, 2.01it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 2.97it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
4/20 3.39G 3.539 5.725 4.275 114 640: 100%|██████████| 16/16 [00:07<00:00, 2.27it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.09it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
5/20 3.47G 3.603 5.681 4.273 145 640: 100%|██████████| 16/16 [00:11<00:00, 1.42it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.16it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
6/20 3.41G 3.672 5.732 4.253 104 640: 100%|██████████| 16/16 [00:11<00:00, 1.41it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.02it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
7/20 3.39G 3.619 5.692 4.263 173 640: 100%|██████████| 16/16 [00:07<00:00, 2.27it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.62it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
8/20 3.39G 3.697 5.732 4.238 69 640: 100%|██████████| 16/16 [00:06<00:00, 2.47it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.45it/s]
all 128 929 0 0 0 0 0
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
9/20 3.39G 3.472 5.697 4.254 71 640: 100%|██████████| 16/16 [00:10<00:00, 1.59it/s]
Class Images Instances Box(P R mAP50 mAP75 mAP50-95): 100%|██████████| 8/8 [00:02<00:00, 3.48it/s]
all 128 929 0 0 0 0 0