border border

Nodes_Auditing Software

Team members

Goh Ray Fong (ESD), Nicholas Tan Yi Da (ESD), Phan Vu Lan (ISTD), Leong Yun Qin Melody (ISTD), Ho Kwan Yee Samuel (ISTD), Koh Chin An Mathias (ISTD)

Instructors:

Matthieu De Mari, Ying Xu, Bikramjit Das

Writing Instructors:

Grace Kong

Teaching Assistant:

Esra Oymak

Project Description

Piki Paki is an automated warehouse solution that aims to make warehouse processes highly efficient.

Through utilising Computer Vision that performs image recognition and counting, this eliminates need for manual and repetitive counting of products to fulfill a particular order

Background

 

  •  1. Rise of E-commerce
  • More SMEs to use warehouse for Fast Moving Consumer Goods (FMCG)

2. Tedious Warehouse Process

Very labour intensive & repetitive warehouses processes occuring

3. Suboptimal Warehouse Efficiency

Processes can be bottlenecks on overall warehouse efficiency which causes

strain on workers & affecting service levels

  • 4. Targetted Bottleneck Process
  • Serious bottleneck at double manual counting of packed orders for verification

 

Piki Paki Benefits 

 

A Computer Vision (CV) application that alleviates the bottleneck process which:

1. Helps to decrease workers' fatigue

2. Helps to increase warehouse processing throughput

 

Piki Paki "4 Step" Process

 

8c281fe3 343d 448a 87f4 39d0259a6aa9

 

1. Order Creation

by manager that creates and sends the delivery orders (DO) and purchase orders (PO)

 

external vision ui and ux xnimrodx lineal xnimrodx 1

 

2. Activate CV Application

by workers who processes order on their workers interface

 

external processing delivery kiranshastry lineal kiranshastry 1

 

3. Automated Data Processing

by CV app that returns item count for worker to verify and submit

external data visualization banking and finance smashingstocks detailed outline smashing stocks 1

 

4. Dashboard Visualisation

on manager interface to provide a clear overview of warehouse processes statuses

Computer Vision in Action!

 

object segmentation 1 1

 

 

1. Object Segmentation

Detects objects & adds bounding boxes to signify product is detected

 

 

 

barcode reading 2

 

 

2. Barcode Detection

Barcodes detected and decoded for product identification, added to product count

 

 

 

aa123771 aca4 411f 8a75 92b1308a489b

 

 

3. Text Detection

If barcode undetected, switch to text detection with Levenshtein Ratio for easier identification

 

 

output 2

 

4. Output

Information grabbed from image sent to server for further processing

 

CV Results

85% ACCURACY!

16ms DETECTION SPEED!

 

Warehouse Storemen Interface

Warehouse workers identify products through a combination of retailer name, PO/DO number and what the product looks like. This interface makes it easy for workers to verify the CV results.

                                

home page google chrome 001

 

 

Homepage

1.

Users can view new and saved orders to process

2.

Search bar to help user in narrowing down orders

                                            3.

New Orders sorted and prioritised

                                               according to deadline starting from

                                               shortest deadline

operation page google chrome 001

******************************************************************************************************

 

 

Order Processing Page

1. 

User uses customised camera app to take image of products and reflect under "Counted" column

 

2. 

"Undo" allows storemen to revert changes made to the item count for each order

 

3. 

"Save" allows user to pause processing an order temporarily, and return later with progress resumed

4. 

"Fill" instantly completes trivially small orders

 

 

Warehouse Manager Interface

Manager Dashboard empowers the manager to create orders, grasp order trends & provides key metrics for the manager to plan ahead to further optimise warehouse processes.

      

0a67263c 8b95 4096 9154 3be92ff03b75

 

 

Dashboard Page

1.

Managers able to see order trend chart for past periods

2.

Able to see tallies of different order statuses

3.

Weekly statistics of order completion rate, incoming order rate as well as average order completion time

4.

These statistics enables manager to easily detect inefficiencies and take action

ordercreation

******************************************************************************************************

 

 

Order Creation Page

1.

Manager is able to create order and add relevant details like items, deadline, invoice ID, customer name

2.

They can also narrow down to see specific orders based on their statuses  at the side bar menu

3.

Item details can also be viewed via the side bar menu.

4.

"Submit" to save and send order to worker interface

                                                             Future Works

 

 

 

 

 

  1.  
  2. 1. Easier Onboarding of more items
  3. No AI training needed when there new items in inventory system, quick scan allow it to be easily detected in future

 

 

 

onboard

 

 

 

  1. 2. Order Picking Recommendation 
  2. Allows storemen fufill orders efficiently, reduce possibility of missed orders that leads to financial losses

 

 

 

document give icon 1

 

 

 

    1.  
    2. 3. Defect Detection on products
    3. Prevents defective goods being packed and shipped to customer, avoiding losses and increase reliability
2247024 200

TEAM MEMBERS

student Goh Ray Fong Engineering Systems and Design
student Nicholas Tan Yi Da Engineering Systems and Design
student Phan Vu Lan Information Systems Technology and Design
student Leong Yun Qin Melody Information Systems Technology and Design
student Ho Kwan Yee Samuel Information Systems Technology and Design
student Koh Chin An Mathias Information Systems Technology and Design
border border