IBM Watson Visual Recognition using image preprocessing

By: Andrew Trice

7   0   1255

Uploaded on 03/08/2017

The Visual-Recognition-Tile-Localization application leverages the Watson Visual Recognition service with image pre-processing techniques to deliver localized image classification. For example, "show me where there is rust on the bridge".

Details and sample code: https://github.com/IBM-Bluemix/Visual-Recognition-Tile-Localization

Bridge Image By Keith Ruffles, CC BY 3.0, https://commons.wikimedia.org/w/index.php?curid=56739084

Comments (5):

By anonymous    2017-09-20

A valid approach to tackle this problem is the use of sliding windows to create smaller subimages out of one image containing multiple entities.

A good example can be found here on GitHub.

Here's a video showing a sliding window approach in action https://youtu.be/wEs1imQmk-Y

Original Thread

By anonymous    2017-09-20

The service works by using a collection of classifiers, each classifier is a single tag only and must be trained with it's own sets of positive and negative images. Try to create smaller sub-images out of one image containing multiple entities.

See one video to verify how it works.

Fork on Github.

Original Thread

By anonymous    2017-09-23

When I use the Visual Recognition, I had the same doubt, after it, I search this article talking about good practices:

The accuracy you will see from your custom classifier depends directly on the quality of the training you perform.

On a basic level, images in training and testing sets should resemble each other. Significant visual differences between training and testing groups will result in poor performance results.

There are a number of additional factors that will impact the quality of your training beyond the resolution of your images. Lighting, angle, focus, color, shape, distance from subject, and presence of other objects in the image will all impact your training. Please note that Watson takes a holistic approach when being trained on each image. While it will evaluate all of the elements listed above, it cannot be tasked to exclusively consider a specific element.

So, the service works by using a collection of classifiers, each classifier is a single tag only and must be trained with it's own sets of positive and negative images. So, the professional recommend using a significantly greater amount of images to improve the performance and accuracy of your classifier such as 100s or 1000s of images.

  • See one video to verify how it works.
  • Fork the Example in the video on Github.
  • Official Documentation about Guidelines for training classifiers.

Original Thread

Popular Videos 335

Submit Your Video

If you have some great dev videos to share, please fill out this form.