Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Yolov8 rejecting labels during model.training #94

Open
palmcorp opened this issue Jun 11, 2024 · 3 comments
Open

Yolov8 rejecting labels during model.training #94

palmcorp opened this issue Jun 11, 2024 · 3 comments

Comments

@palmcorp
Copy link

CASE: Yolov8-pose rejecting labels during training
QUESTION: Where does yolov8 determine how many columns are required for each label?

ERROR at model.train phase:

train: WARNING ⚠️ /content/datasets/echo-pose/images/train2017/frame_000151.PNG: ignoring corrupt image/label: labels require 15 columns each

SAMPLE LABELS FROM train2017:
0 0.680863 0.595223 0.20369 0.0726488 0.579017 0.631547 2 0.687827 0.609047 2 0.748973 0.590758 2 0.782708 0.558898 2 0.6325 0.624151 2

0 0.659062 0.613906 0.23378 0.0539435 0.542172 0.640877 2 0.645104 0.61308 2 0.714791 0.617574 2 0.775952 0.586934 2 0.589449 0.637366 2
<etc.>

BACKROUND

  1. no similar problems weere found in the GitHub blog
  2. Label files were scanned to insure no hidden, or unicode characters were embedded

#####C O D E#####

LOAD MODEL

ECHO_YAMAL=os.path.join ("/content/YAML/yolov8n-echo-pose.yaml")
model = YOLO(ECHO_YAMAL, verbose = VERBOSE)
model = YOLO('yolov8n-pose.pt') # load a pretrained model (recommended for training)
model = YOLO('yolov8n-pose.yaml').load('yolov8n-pose.pt')

TRAIN MODEL

YP=os.path.join ("/content/YAML/data5.yaml")
model.train(data=YP, epochs=EPOCHS, imgsz=640)

HERE ARE THE TWO YAML FILES:

=====================
yolovn-echo-pose.yaml

Ultralytics YOLO 🚀, AGPL-3.0 license

YOLOv8-pose-p6 keypoints/pose estimation model. For Usage examples see https://docs.ultralytics.com/tasks/pose

Parameters

nc: 5 # number of classes # UPDATED *
kpt_shape: [5, 2] # number of keypoints,* number of dims (2 for x,y or 3 for x,y,visible)
flip_indx: [0,1,2,3,4] # no flipping *
scales: # model compound scaling constants, i.e. 'model=yolov8n-p6.yaml' will call yolov8-p6.yaml with scale 'n'

[depth, width, max_channels]

n: [0.33, 0.25, 1024]
s: [0.33, 0.50, 1024]
m: [0.67, 0.75, 768]
l: [1.00, 1.00, 512]
x: [1.00, 1.25, 512]

YOLOv8.0x6 backbone

backbone:

[from, repeats, module, args]

  • [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
  • [-1, 1, Conv, [128, 3, 2]] # 1-P2/4
  • [-1, 3, C2f, [128, True]]
  • [-1, 1, Conv, [256, 3, 2]] # 3-P3/8
  • [-1, 6, C2f, [256, True]]
  • [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
  • [-1, 6, C2f, [512, True]]
  • [-1, 1, Conv, [768, 3, 2]] # 7-P5/32
  • [-1, 3, C2f, [768, True]]
  • [-1, 1, Conv, [1024, 3, 2]] # 9-P6/64
  • [-1, 3, C2f, [1024, True]]
  • [-1, 1, SPPF, [1024, 5]] # 11

YOLOv8.0x6 head

head:

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 8], 1, Concat, [1]] # cat backbone P5

  • [-1, 3, C2, [768, False]] # 14

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 6], 1, Concat, [1]] # cat backbone P4

  • [-1, 3, C2, [512, False]] # 17

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 4], 1, Concat, [1]] # cat backbone P3

  • [-1, 3, C2, [256, False]] # 20 (P3/8-small)

  • [-1, 1, Conv, [256, 3, 2]]

  • [[-1, 17], 1, Concat, [1]] # cat head P4

  • [-1, 3, C2, [512, False]] # 23 (P4/16-medium)

  • [-1, 1, Conv, [512, 3, 2]]

  • [[-1, 14], 1, Concat, [1]] # cat head P5

  • [-1, 3, C2, [768, False]] # 26 (P5/32-large)

  • [-1, 1, Conv, [768, 3, 2]]

  • [[-1, 11], 1, Concat, [1]] # cat head P6

  • [-1, 3, C2, [1024, False]] # 29 (P6/64-xlarge)

  • [[20, 23, 26, 29], 1, Pose, [nc, kpt_shape]] # Pose(P3, P4, P5, P6)

Classes

names:
0: Class0
1: Class1
2: Class2
3: Class3
4: Class4

==========
data5.yaml

#keypoint data sets

Data

#path: D:POSE/data
path: /content/datasets/echo-pose
train: images/train2017 # subsets to D
val: images/val2017 # relative to path

Keypoints

kpt_shape: [5,2] # number of keypoints, number of dim
flip_indx: [0,1,2,3,4] # no flipping

#backbone: uskng standard backbone
backbone:

[from, repeats, module, args]

  • [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
  • [-1, 1, Conv, [128, 3, 2]] # 1-P2/4
  • [-1, 3, C2f, [128, True]]
  • [-1, 1, Conv, [256, 3, 2]] # 3-P3/8
  • [-1, 6, C2f, [256, True]]
  • [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
  • [-1, 6, C2f, [512, True]]
  • [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32
  • [-1, 3, C2f, [1024, True]]
  • [-1, 1, SPPF, [1024, 5]] # 9
    #usn standard head

head:

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 8], 1, Concat, [1]] # cat backbone P5

  • [-1, 3, C2, [768, False]] # 14

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 6], 1, Concat, [1]] # cat backbone P4

  • [-1, 3, C2, [512, False]] # 17

  • [-1, 1, nn.Upsample, [None, 2, "nearest"]]

  • [[-1, 4], 1, Concat, [1]] # cat backbone P3

  • [-1, 3, C2, [256, False]] # 20 (P3/8-small)

  • [-1, 1, Conv, [256, 3, 2]]

  • [[-1, 17], 1, Concat, [1]] # cat head P4

  • [-1, 3, C2, [512, False]] # 23 (P4/16-medium)

  • [-1, 1, Conv, [512, 3, 2]]

  • [[-1, 14], 1, Concat, [1]] # cat head P5

  • [-1, 3, C2, [768, False]] # 26 (P5/32-large)

  • [-1, 1, Conv, [768, 3, 2]]

  • [[-1, 11], 1, Concat, [1]] # cat head P6

  • [-1, 3, C2, [1024, False]] # 29 (P6/64-xlarge)

  • [[20, 23, 26, 29], 1, Pose, [nc, kpt_shape]] # Pose(P3, P4, P5, P6)

Classes

names:
0: Class0
1: Class1
2: Class2
3: Class3
4: Class4
@pderrenger
Copy link
Member

Hello!

Thank you for providing detailed information about your issue. It looks like you're encountering a problem with label format during the training phase of your YOLOv8-pose model. Let's address your concerns step-by-step.

Label Format Issue

The error message indicates that your labels require 15 columns each, but the provided labels do not match this requirement. For pose estimation, the label format should follow the structure outlined in the Ultralytics documentation:

Format with Dim = 2:

<class-index> <x> <y> <width> <height> <px1> <py1> <px2> <py2> ... <pxn> <pyn>

Format with Dim = 3:

<class-index> <x> <y> <width> <height> <px1> <py1> <p1-visibility> <px2> <py2> <p2-visibility> ... <pxn> <pyn> <pn-visibility>

Given your kpt_shape: [5, 2], each label should have:

  • 1 class index
  • 4 bounding box coordinates (x, y, width, height)
  • 5 keypoints, each with 2 coordinates (px, py)

This totals to (1 + 4 + (5 \times 2) = 15) columns per label.

Sample Label Correction

Your sample labels should look like this:

0 0.680863 0.595223 0.20369 0.0726488 0.579017 0.631547 0.687827 0.609047 0.748973 0.590758 0.782708 0.558898 0.6325 0.624151

YAML Configuration

Ensure your YAML files correctly reflect the dataset structure and keypoints configuration. Here’s a simplified version of your data5.yaml:

# Data
path: /content/datasets/echo-pose
train: images/train2017
val: images/val2017

# Keypoints
kpt_shape: [5, 2]  # number of keypoints, number of dims (2 for x,y or 3 for x,y,visible)
flip_idx: [0, 1, 2, 3, 4]  # no flipping

# Classes
names:
  0: Class0
  1: Class1
  2: Class2
  3: Class3
  4: Class4

Next Steps

  1. Verify Label Format: Ensure all your label files conform to the required format.
  2. Update Packages: Make sure you are using the latest versions of torch and ultralytics. You can update them using: 
    pip install --upgrade torch ultralytics
  3. Minimum Reproducible Example: If the issue persists, please provide a minimum reproducible code example. This helps us investigate the problem more effectively. You can find guidelines here.

Example Code

Here’s a snippet to help you get started with training after ensuring your labels and configurations are correct:

from ultralytics import YOLO

# Load a pretrained model
model = YOLO('yolov8n-pose.pt')

# Train the model
results = model.train(data='/content/YAML/data5.yaml', epochs=100, imgsz=640)

Feel free to reach out if you have any more questions or need further assistance. Happy training! 🚀

@palmcorp
Copy link
Author

palmcorp commented Jun 11, 2024 via email

@pderrenger
Copy link
Member

Hi Paul,

Thank you for your kind words and for your continued efforts in resolving this issue. 😊

To ensure we can assist you effectively, please verify a couple of things:

  1. Label Format: Double-check that all your label files strictly follow the required format for pose estimation. Each label should have 15 columns, as outlined previously.

  2. Package Versions: Ensure you are using the latest versions of torch and ultralytics. You can update them with:

    pip install --upgrade torch ultralytics

  3. Minimum Reproducible Example: If the issue persists, could you please provide a minimum reproducible code example? This will help us investigate the problem more effectively. You can find guidelines on how to create one here.

Here’s a quick example to get you started with training after ensuring your labels and configurations are correct:

from ultralytics import YOLO

# Load a pretrained model
model = YOLO('yolov8n-pose.pt')

# Train the model
results = model.train(data='/content/YAML/data5.yaml', epochs=100, imgsz=640)

Feel free to reach out if you have any more questions or need further assistance. The YOLO community and the Ultralytics team are here to help!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@palmcorp @pderrenger