What is Semantic Segmentation, Instance Segmentation, Panoramic segmentation?
Data Labeling Service In Self-driving and Medical Industry
Image segmentation is a very important research and application direction in computer vision. According to certain rules, images are divided into different parts and labeled at pixel level.
Image Segmentation
The diagram is as follows:
1. Image classification
Identify the contents of the image, as shown in the following figure: person, tree, grass, sky
2. Object detection
Identify the content and its position in the image, as shown in the following figure: the person
3. Semantic segmentation
Each pixel is labeled in the image in categories. As shown in the following figure, the image is divided into human (red), tree (dark green), grass (light green), sky (blue).
4. Instance segmentation
The combination annotation of target detection and semantic segmentation. The target detection comes first, and then each pixel is labeled (semantic segmentation). Compared to the image above, we take the person as the target objection for example:
Semantic segmentation does not distinguish different instances in the same category (all people are marked red).
Instance segmentation distinguishes different instances of the same category. (different people are distinguished by different colors)
5. Panoramic segmentation
Another combination annotation of semantic segmentation and instance segmentation, more complex. Panoramic segmentation means that all targets should be detected and different instances in the same category should be distinguished. Compared to the image above:
Instance segmentation only detects and divides the objects in the image (such as the people in the above figure) and distinguishes them using different colors.
Panoramic segmentation is to detect and segment all objects in the picture, including the background, and distinguish different instances (different colors are used).
Mainly Application
In the process of autonomous driving, the car itself needs to have a number of “skills” such as perception, planning, decision-making, and control, which can be collectively referred to as “artificial intelligence”.
Semantic segmentation provides information about the free space on the road, as well as the detection of landmark and traffic signs.
The result of semantic segmentation is to transform the image into several color blocks, and each color block represents one part of the image.
End
Outsource your data labeling tasks to ByteBridge, you can get the high-quality ML training datasets cheaper and faster!
- Free Trial Without Credit Card: you can get your sample result in a fast turnaround, check the output, and give feedback directly to our project manager.
- 100% Human Validated
- Transparent & Standard Pricing: clear pricing is available(labor cost included)
Why not have a try?
source:
https://my.oschina.net/u/876354/blog/3055850
Author:雪饼（BigdataAILab）
Relevant Articles:
2 How Auto-Driving Achieved through Machine Learning?
3 Labeling Service Case Study — Video Annotation — License Plate Recognition
4 High-Quality Training Data for Autonomous Cars in 2021
5 How Data Labeling and Annotation Services Empower Logistics in 2021?
