EZDLCustom Image Recognition

Build custom machine learning models without having to code

Start training
Read document
FAQ
Custom Models
Image Classification
Classify the object/scene shown in an image, used when there is only one object/scene shown in the image
Dining roomBathroom
Fast model training with small dataset
Training set requires only 20-100 images assigned to each label. Training time can be as short as 10 minutes.
Diversified applications
Image content identification: custom identification, showing more business-specific image recognition results
Industrial sorting or quality inspection: custom conveyor belt product identification, as the basis of realizing automatic sorting or quality inspection
Image censoring: custom image censoring rules, such as identifying smoking during live broadcast
Others: just brainstorm and train the image classification model you desire
Object Detection
Identify the label and position of each target object in the image and count the number of objects by label,
suitable for identifying multiple (types of) objects in the image.
VaseCup
Fast model training with small dataset
Training set requires only 20-100 images assigned to each label. Training time may take more than 30 minutes.
Diversified applications
Object counting: such as counting goods on shelves for retail business or counting cells for medical research
Industrial quality inspection: such as detecting positions of minor product defects and count the total number of defects
Others: just brainstorm and train the image classification model you desire
Steps to Use the Platform
4 steps only
steps
1.Create a model
2.Upload and label images/objects
3.Train and test the model
4.Get the model deployed with cloud API or offline SDK
How to Integrate after Training the Model
Cloud API
It provides enterprise-level stable and accurate high-flow services with millisecond-level response time, flexible high concurrency handling, and 99% reliability.
Offline SDK
It provides client SDK, supporting offline running in operating systems including iOS and Android. It can be integrated in mobile phones, cameras and other terminal equipment.
Core Advantages
Easy-to-use
Simple steps to use
4 steps only, zero knowledge of deep learning required
Fast
Optimized algorithms and engineering, <15 minutes of training time on average
High-performance
More than 2/3 of models
get accuracy scores >90%
Highly secure
Data encrypted and isolated,
mature authentication mechanism
Image Classification
Classify the object/scene shown in an image, used when there is only one object/scene shown in the image.
Image Classification
Object Detection
Identify the label and position of each target object in the image, suitable for identifying multiple objects in the image.
Object Detection

Contact Us

Industry:
Click to refresh

Contact Us

Success

Your requirements have been received, we will contact you soon

Please follow our official Wechat account to keep in pace with our latest news!

Contact Us

出了点小问题,请稍候重试

百度AI微信公众号