πŸŽ„ 2024 Advent Challenge 4: Unwrap the Magic of Object Detection 🎁

News Challenges
5

Hi @pradeep_ch , This one was really fascinating to work with. Few months back this capability seemed to be a long shot for low code solutions.

Awesome to see UiPath making it accessible to all aspirants.

Here is my final submission for Challenge Number 4.

  1. Here is the activity utilized for the this task.
    The image shows a UI interface of the "MLSkill" activity configuration, highlighting an object detection ML skill with input and output types and a section for deserializing JSON in a workflow. (Captioned by AI)
    The image shows a UI interface of the "MLSkill" activity configuration, highlighting an object detection ML skill with input and output types and a section for deserializing JSON in a workflow. (Captioned by AI)379Γ—522 7.28 KB
  2. Folders used to process & store output
    The image shows two folders, "For RPA Bot" and "OutputFolder," with a description indicating that files are read from the "For RPA Bot" folder and saved to the "Output Folder" after processing. (Captioned by AI)
    The image shows two folders, "For RPA Bot" and "OutputFolder," with a description indicating that files are read from the "For RPA Bot" folder and saved to the "Output Folder" after processing. (Captioned by AI)285Γ—478 26.7 KB
  3. Following are the results for varying input categories with ranging backgrounds.

A)A family is shopping in a grocery store, and an object detection model incorrectly identifies a person as a dog with a confidence level of 51%.

A family is shopping in a grocery store, and an object detection model incorrectly identifies a person as a dog with a confidence level of 51%. (Captioned by AI)
A family is shopping in a grocery store, and an object detection model incorrectly identifies a person as a dog with a confidence level of 51%. (Captioned by AI)1362Γ—759 183 KB

B) Here object detection model has labeled two people and a dog, with one misclassification identifying the dog as a β€œteddy bear” with 66% confidence.

The screenshot shows a UIPath environment where an object detection model has labeled two people and a dog, with one misclassification identifying the dog as a "teddy bear" with 66% confidence. (Captioned by AI)
The screenshot shows a UIPath environment where an object detection model has labeled two people and a dog, with one misclassification identifying the dog as a "teddy bear" with 66% confidence. (Captioned by AI)1068Γ—662 86.3 KB

C) Shows that the Yolo model is pretty well optimized for traffic scenarios.

This image shows a screenshot of a machine learning model's performance in identifying and categorizing various types of vehicles with confidence scores on a busy road, highlighted by bounding boxes. (Captioned by AI)
This image shows a screenshot of a machine learning model's performance in identifying and categorizing various types of vehicles with confidence scores on a busy road, highlighted by bounding boxes. (Captioned by AI)1067Γ—618 105 KB

That was all. Once again thanks all the seniors for this opportunity. Ya’ll have an awesome holiday season.
Project -
AdventC4_HS.zip (8.5 KB)

Cheers !

4 Likes