361 questions
Advice
0
votes
0
replies
43
views
Requesting an opinion about an extreme optimization pipeline for a YOLOv8 model
i have an idea about an optimization method that i think if it is done right, it could result for an extremely light model. The Method evolves around a multi-step methodology that either reduce the ...
Advice
0
votes
0
replies
37
views
YOLOv8/v11: How to force 4-channel (NxNx4) input without the trainer overriding 'ch' back to 3?
I am trying to train a YOLOv11 model (Ultralytics) on a custom dataset that uses 4-channel inputs (RGB + a custom metadata channel).
I’ve modified my architecture YAML to set ch: 4, and I’m ...
- 497
Advice
0
votes
1
replies
80
views
Tracking small targets on thermal video
I'm tracking a target from an UAV using a thermal camera. Detection is YOLOv8n running every N frames on an NPU, and I need something to hold the track between detections.
What I've tried:
Template ...
- 1
Advice
2
votes
3
replies
128
views
How to analyze classroom behavior using computer vision and pose estimation?
I am trying to build a computer vision system to analyze classroom behavior from surveillance cameras.
The goal is to automatically detect several behavioral indicators such as:
- student attention
- ...
- 1
0
votes
1
answer
3k
views
OSError: [WinError 1114] A dynamic link library (DLL) initialization routine failed when importing torch on Windows
I am getting the following error when trying to use PyTorch-based libraries (YOLO / ultralytics) on Windows. The error occurs on import torch, so any torch-based library fails.
System information
OS: ...
0
votes
1
answer
147
views
Fail to Run OnnxRuntime Session in C# with CUDA Device
I created a YOLO v11 DL model using a given dataset and both the train and prediction modules work fine in PyCharm.
However, after exporting the model file to ONNX format and running the prediction ...
1
vote
0
answers
42
views
Why does the same YOLOv8n-pose model with different weights have significantly different inference speeds?
I’m testing YOLOv8n-pose models that share the exact same architecture, input size, hardware (GPU), framework, batch size, and precision settings. The only difference between them is the trained ...
- 11
0
votes
1
answer
109
views
Cannot calculate confusion matrix utilizing supervision from roboflow for Yolov8 model
I am trying to calculate the confusion matrix for my yolov8 (or yolov11) model utilizing supervision from roboflow. I found some instructions but they do not seem to be crystal clear. For example ...
- 5,747
0
votes
0
answers
63
views
Error while converting quantized Torch model to ONNX
I’m applying QAT to YOLOv8n model with the following configuration:
QConfig(
activation=FakeQuantize.with_args(
observer=MovingAverageMinMaxObserver,
quant_min=0,
quant_max=...
- 111
0
votes
0
answers
17
views
Yoliv8 instance segmentation works on Collab but not on windows pc
I am doing a yolov8 instance segmentation project. I creates a dataset with a few images using roboflow
When I run it on my windows pc, it keeps saying the images are corrupted.
But I have no issue ...
0
votes
1
answer
134
views
i want to custmize data.yaml file for train yolo11x.pt model by ultralytics?
class_id x_center y_center width height behavior_id
eg.txt file data 6 0.260313 0.739167 0.131875 0.038333 1
6 0.580313 0.821250 0.290625 0.245834 0
6 0.821562 0.775416 0.230625 0.179167 0
6 0.914062 ...
- 1
0
votes
0
answers
62
views
Edit PYTHONPATH within Docker
I'm trying to run a YOLO-World V2 demo script inside a Docker/Buildah container, but I’m getting an import error related to a custom module:
ImportError: Failed to import custom modules from {'imports'...
- 1
-1
votes
2
answers
169
views
YOLOv8 Custom Model Fails to Detect Card Elements — What Am I Doing Wrong?
I’m working on a card verification system where the goal is to detect fixed visual elements on a card (e.g., logo, photo, chip, signature) using computer vision.
Given a photo of a card, I want to ...
- 67
0
votes
0
answers
197
views
How do I deploy local model weights (best.pt) to CVAT for auto-annotation
I'm trying to use CVAT locally with Docker for image bounding box labeling. To save time, I'd like to use auto-annotation with a partially trained model. However, when I try to deploy the model from ...
1
vote
0
answers
264
views
yolov8n onnx for object detection (ultralytics library) -> how to postprocess the output?
When using the Ultralytics library to perform model inference directly on an image, the model correctly predicts the bounding box positions and class identities:
!pip install ultralytics opencv-python
...
- 39