Computer VisionA fast growing branch of computer science thanks to recent progress in ML. Classic tasks in ML such as Object Detection or Image Segmentation extended the definition of Computer Vision well beyond the simple graphic algorithms or optic solutions.
|
|
Object Detection
Object Detection is an important task in Computer Vision. With object detection models you can infer what objects exist in a given image and where exactly they are located. There is a label assigned to each object detection instance along with a confidence score which shows how confident the model was in detecting the object.
This object detection demo calls a running API that can detect two classes of sofa and picture. I trained this model on a few images and the accuracy might not the best. Accuracy and performance are limited due to the resource limitations of this demo.
Lower "confidence level" to detect less confident objects. Use "Clear" to reset and upload another image.
By the way, those are my own paintings in the example picture!
Demo only, limited accuracy and performance!
This object detection demo calls a running API that can detect two classes of sofa and picture. I trained this model on a few images and the accuracy might not the best. Accuracy and performance are limited due to the resource limitations of this demo.
Lower "confidence level" to detect less confident objects. Use "Clear" to reset and upload another image.
By the way, those are my own paintings in the example picture!
Demo only, limited accuracy and performance!
Instance Segmentation
Instance Segmentation is a sub-task of image segmentation with variety of applications. Image segmentation performs object detection per pixel and masks pixels that belong to an object. Instance segmentation is a specialized algorithm to detect instances of a specific object. In the example below, the model only detects pixels related to one category of objects, i.e. wires or cables.
Simply take a photo and upload it here or just use the example images below.
Demo only, limited accuracy and performance!
Simply take a photo and upload it here or just use the example images below.
Demo only, limited accuracy and performance!
Technical Description
YoloV8 image segmentation model trained on a small dataset.
Updates in demo V3:
Updates in demo V3:
- Fixed fusing bug
- Retrained with more efficient data augmentation
|
Picture SwapTry this Computer Vision demo to replace a picture with another using image segmentation model. |
