Ultralytics
yolov5/tutorials/train_custom_data/ #8456
giscus[bot]
bot
Replies: 14 comments 17 replies
-
|
can i work with labels personnalized for exemple i work with invoice object detection so labels are invoice_no, .... ? |
Beta Was this translation helpful? Give feedback.
-
|
Absolutely, you can work with customized labels like Here's a quick guide to get you started:
Remember, the key to a successful object detection model is a well-labeled, diverse dataset. So, take your time with the data preparation step. Happy detecting! 🎉 |
Beta Was this translation helpful? Give feedback.
-
|
I am more familiar to yolov8 but have to work with and am new to yolov5, but I cannot get a custom v5 model to run using a python command like model = YOLO('./model.pt') . How can I use this model to get bounding box coordinates of objects in the images? |
Beta Was this translation helpful? Give feedback.
-
|
import torch img = '/content/test/images/20190925810_jpg.rf.f9e58a3b226d7f3f00366d9531a41e6a.jpg' results = model(img) u can use sth like that it works for me |
Beta Was this translation helpful? Give feedback.
-
|
Hey there! 😊 It looks like you're trying to use a YOLOv5 model with a syntax that's more aligned with YOLOv8. For YOLOv5, you'll want to use the import torch
from PIL import Image
# Load your custom model
model = torch.load('./model.pt')
model.eval() # Set the model to evaluation mode
# Load your image
img = Image.open('path/to/your/image.jpg')
# Perform inference
results = model(img)
# Extract bounding box coordinates
for *xyxy, conf, cls in results.xyxy[0]:
print(f'Coordinates: {xyxy} | Confidence: {conf} | Class: {cls}')This should get you started with running your YOLOv5 model on your images. If you have any more questions or need further assistance, feel free to ask. Happy coding! 🚀 |
Beta Was this translation helpful? Give feedback.
-
|
How to enable/disable augmentations when training custom YOLOv5 model from launching |
Beta Was this translation helpful? Give feedback.
-
|
I guess it is |
Beta Was this translation helpful? Give feedback.
-
|
Is it possible to use the content provided in this repository (specifically, the pretrained model table types mentioned in point #3) for a research paper, ensuring proper citation? |
Beta Was this translation helpful? Give feedback.
-
|
@AsmaMubarakk hello! Absolutely, you are welcome to use the pretrained model table types mentioned in our repository for your research paper. We appreciate you ensuring proper citation 📚. In citing our work, you might refer to the specific GitHub repo URL and indicate it was accessed on [date of access]. If there's a related publication or documentation, citing that as well would be beneficial. For instance: If you have any more questions or need further details for your citation, feel free to ask. Happy researching! 🚀 |
Beta Was this translation helpful? Give feedback.
-
|
Hi, is it possible to create a YOLOv5 dataset from a CSV-file? |
Beta Was this translation helpful? Give feedback.
-
|
Hello, yolov5 accepts datasets that contain images in one folder and their annotations in another folder so ur CSV file contains for example the path of your image and the annotation of each class, |
Beta Was this translation helpful? Give feedback.
-
|
Hello! 😊 Yes, you can certainly create a YOLOv5 dataset from a CSV file. Here's a quick guide on how to do it:
Here's a tiny bit of Python pseudocode to get you started: import pandas as pd
df = pd.read_csv("data.csv")
for index, row in df.iterrows():
# Process your rows to YOLO format
with open(f"{row['image_path'][:-4]}.txt", 'w') as file:
file.write(f"{row['class']} {row['x_center']} {row['y_center']} {row['width']} {row['height']}\n")Just replace the column names with ones corresponding to your CSV. Let me know if there's anything else you'd like to know! |
Beta Was this translation helpful? Give feedback.
-
|
I am facing this issue "ValueError: malformed node or string: <ast.Name object at 0x0000010EFA2BE550>" when i convert trained yolov5 on custom data and convert into tflite after that add metadata file then I test on image the tflite . I am facing this isssue. I have only one class "ball". |
Beta Was this translation helpful? Give feedback.
-
|
@Muhammad-Irfan-Shahzad it seems like you're encountering an issue with the conversion of a YOLOv5 model to TensorFlow Lite (TFLite) format and adding metadata. This error typically arises from a problem in the script or the way the model is being handled during conversion or testing. To assist you better, could you please share the specific code snippet you're using for the conversion to TFLite and the addition of metadata? This will help in pinpointing the exact cause of the error. Meanwhile, ensure that your environment is set up correctly with all necessary dependencies, and that the model and data paths are correctly specified. Here’s a basic example of how you might convert a YOLOv5 model to TFLite: import torch
from yolov5 import YOLO
# Load the trained YOLOv5 model
model = YOLO('path/to/your/custom_trained_yolov5_model.pt').autoshape() # for file/URI/PIL/cv2/np inputs and NMS
model.eval()
# Convert to TorchScript
ts_model = model.model.to_torchscript()
# Save the TorchScript model
ts_model.save('model.torchscript.pt')
# Convert TorchScript model to TensorFlow Lite
converter = torch.jit.load('model.torchscript.pt')
converter = torch.jit.optimize_for_mobile(converter)
converter._save_for_lite_interpreter('model.tflite')Make sure to adjust the paths and model names as per your setup. If you continue to face issues, please provide the error logs and code snippet for more detailed guidance. |
Beta Was this translation helpful? Give feedback.
-
|
i am facing the issue in my custom model training i have tried more then 10 times to train the model but in 8 time it is showing corrupt images as error how can i fix it. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, It seems like you're encountering issues with corrupt images during your training process. Here are a few steps to help resolve this:
For the AMP issue, it's disabled due to a check failure, which might be related to the corrupt images or specific configurations in your setup. Once the image issues are resolved, AMP might work as expected. If the problem persists after trying these steps, please provide more details about your setup and any error logs for further assistance. |
Beta Was this translation helpful? Give feedback.
-
|
i am facing the issue in my custom model training i have tried more then 10 times to train the model but in 8 time it is showing corrupt images as error how can i fix it. |
Beta Was this translation helpful? Give feedback.
-
|
It looks like you're encountering issues with corrupt images during your training process. Here are a few steps you can take to resolve this:
If the problem persists, consider providing more details or sharing your dataset configuration for further assistance. |
Beta Was this translation helpful? Give feedback.
-
|
Dear developers, I would like to ask, I have my own training set, but every training results are different, whether using pre-training weights or not, I would like to ask if there is any way to ensure that the results can be repeated under the same configuration, thank you very much, this problem has troubled me for a long time. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, Thank you for reaching out with your query about achieving consistent training results with YOLOv8. To ensure reproducibility in your model training, you can follow these steps:
By following these steps, you should be able to see more consistent results across your training runs. If you continue to experience variability, it might be useful to check if any part of your data pipeline or model configuration changes between runs. For more detailed guidance, please refer to our documentation on training custom data. |
Beta Was this translation helpful? Give feedback.
-
|
Hello all, can anyone point me in the direction of the documentation explaining the column titles in the results.csv file, specifically the one generated by YOLOv10, but any explanation will be greatly appreciated. |
Beta Was this translation helpful? Give feedback.
-
|
@krausekeiii hello! Thank you for reaching out. The
These columns provide a comprehensive view of the model's performance and training progress. For a more detailed explanation of each metric and how they are calculated, you can refer to the Ultralytics documentation. If you have any specific questions about the columns or need further clarification, feel free to ask! 😊 |
Beta Was this translation helpful? Give feedback.
-
|
Dear developers, I would like to ask for help training a Yolov7 model for plant disease detection...as I am new to it I do not know how to start even from the data loading up to saving the model....how to show the confusion matrix , graphs, and all...I don't how to train a model... I know only up to annotation. |
Beta Was this translation helpful? Give feedback.
-
|
@Syed-bisma2 to get started with training a YOLOv7 model for plant disease detection, you'll need to prepare your dataset in the YOLO format, train the model using the Ultralytics library, and visualize results like confusion matrices. You can find a detailed guide on training custom data in the Ultralytics documentation. If you have specific questions, feel free to ask here. |
Beta Was this translation helpful? Give feedback.
-
|
I've got some custom splits (eg. I'm trying to quantize my model, so I've got splits named quantize, finetune, eval). I want to use yolov5 val.py for this. Can I have such custom split names in my dataset yaml? How can I make sure val.py uses the right split? |
Beta Was this translation helpful? Give feedback.
-
|
Seems like check_dataset is only looking for specific train, val, test, download |
Beta Was this translation helpful? Give feedback.
-
|
@NikilXYZ the |
Beta Was this translation helpful? Give feedback.
-
|
my data is in .npy grayscale format i want to change the data loading pipeline to adjust to .npy file but i am not able to locate where to find the data loading part from the file paths |
Beta Was this translation helpful? Give feedback.
-
|
@AashishThakur1321 to modify the data loading pipeline for |
Beta Was this translation helpful? Give feedback.
-
|
i have trained my model to detect objects but it is not able to detect small object, how can i improve my accuracy to detect small objects? |
Beta Was this translation helpful? Give feedback.
-
|
@AashishThakur1321 to improve the detection of small objects, you can try increasing the input image size during training (e.g., |
Beta Was this translation helpful? Give feedback.
-
yolov5/tutorials/train_custom_data/
Learn how to train your data on custom datasets using YOLOv5. Simple and updated guide on collection and organization of images, labelling, model training and deployment.
https://docs.ultralytics.com/yolov5/tutorials/train_custom_data/
Beta Was this translation helpful? Give feedback.
All reactions