Bullet Fault Detection System using Real time image processing

Project Title

Bullet Fault Detection System using Real time image processing

Project Area of Specialization Artificial IntelligenceProject Summary

Introduction :

Bullet Fault Detection Using Real-time image processing and Machine Learning is a system which can be implemented in final phases of the bullet manufacturing industries .This system can check and analyse the final output of the bullet manufacturing industry, as in this project the product which will be detected and analyzed are the cases of the bullets.

Abstract  :

Real time monitoring of bullets on a conveyer belt and repair the known faults(water marks, scratch, perforation etc), through the two cameras attached on the top of the system, they will monitor the faulty bullets and separate them by using YOLO V3(You Only Look Once – Version 3) for image processing.

Project Objectives

- Software & Libraries Installation ?
- Literature Review of Networks ( Like Neural , Coco Inception )                             ?
- Training Files & Yolo Weights    ?
- Bullet Samples ( By POF )           ?
- Bullet Images ( 2500 Images )      ?
- Image Annotation ( 700 )              ?
- Pre-Trained Model ( Detecting Real time objects like Book , Bottle , Person etc ) ?

 - Real Time Detection of Playing Cards through image processing & Machine Learning      ?  
    Detection of Bullet                             ?

    Camera Integration & synchronization
    Detecting Faults

    Conveyer Belt
    Testing of Actual Hardware
     
   
  All tick marked are done . 
    Project Implementation Method

Benefits of the Project

Scope of the Project and Quantifiable Outcomes :

       - To Reduce labor cost in industry thus increasing the                 level of profit.
 

• Production Quantity improves significantly as there is no manual handling and everything is done on Automated system.
   
• Time is saved as the work is being executed with fully computerized operation and comprehensive data processing facilities.
   

           To improve efficiency.

Technical Details of Final Deliverable

• - Bullet Fault Detection on a real-time video .
  
  - Running yolo v3 on pre trained data.

- Data training on yolo v3.

- Acquiring dataset for bullet fault detection

- Grey Scaled Data Training

- Training yolo model for bullet fault detection.

- Hardware Development

- Bounding Box Prediction

Following YOLOv3 our system predicts bounding boxes using dimension clusters as anchor boxes [15]. The network predicts 4 coordinates for each bounding box, tx, ty, tw, th. If the cell is offset from the top left corner of the image by (cx, cy) and the boundingbox prior has width and height pw, ph, then predicts .

The image Below shows the Real-time prediction of bullet not the pre-trained ( saved data comparison ) .

Final Deliverable of the Project HW/SW integrated systemType of Industry Manufacturing Technologies Artificial Intelligence(AI)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	73000
High Resolution Camera	Equipment	2	7000	14000
High end Development Pc	Equipment	1	55000	55000
Spiral Rod Conveyer	Miscellaneous	1	4000	4000