Water Related

sustainable fishery

AP014

snehil jayswal (nirma university)

Aug 10, 2021 631 views

sustainable fishery

Our venture is coming up with the cutting edge End-to-End product which can help the marine species and over a 5-10 years course wild capture would be rejuvenated naturally with the ultimate solution what we offer with the existing Hardware/Software but integrating and applying it for a unique way.
Blind Fishing and overfishing has made the marine resources / wild capture as no longer a bottom less fishing.
This overfishing put a trouble to 1/3 of world population especially the under-developed and developing countries who rely ocean as their cheap protein.

Project Proposal


1. High-level project introduction and performance expectation

Our team is coming up with the cutting edge End-to-End product which can help the marine species and over a 5-10 years course wild capture would be rejuvenated naturally with the ultimate solution what we offer with the existing Hardware/Software but integrating and applying it for a unique way.

 

 

 

Our Tier-2 Customer base is 1.3 million fishing vessels among the 4 million in world wide.

Our Tier-1 Customer base is 400,000 fishing vessels which are > 100 tons catching capacity. (goes to 500,000 metric ton processing of catchment in a single vessel in an year)

So government and government subsidies enrolled programs.

Private parties such as Fishing vessel builders of Mechanical vessel to Fully automated processing plant in fishing vessel.

Throughout the worlds - fishing subsidy per year is ~ US $ 15 billion per year.

Many countries and builders planned to deploy solution by 2025/30

 

X 
Heading 1 
Hoa&ng 
IOTG Workbook • 
Status

 

Incubation

De-weeding (external fishes out from Native breed)

Diseasies / health department

Edge.AI

Phase0: No PC / No Edge.AI / No IP Cams (Ideation)

Feed offline datas & train the system and create your trained database

Phase1: Incubation

Edge.AI --> $2000

Internal GPU (TGL / ADL --> GPU as part of internal)-

/Discrete GPU (Nvidia, Intel DG1/DG2)

IP POE Cams --> $300 - 600 (1 to start)

Phase2: performing / scaling

Intel FPGA --> AI activities --> GPU EU (Execution Units) --> Meta Data --> GPU-

 

2. Block Diagram

3. Expected sustainability results, projected resource savings

Why only intel GPU is good compared to other competitors ?(acc to us, following are few reasons):

Using Intel Core i7-3770 CPU (4 cores, 8 threads) [1]  , it has got 74% accuracy of  even though this intel core is low configuration, if we use intel GPU or FPGA(agilex which is ai based) , we can surely Reach precision/accuracy of 90% or more.

And by using Four NVIDIA V100 GPU  (high end PC)- accuracy is 83 % [2]


 Role of FPGA:

With hep of intel GPU, we are sure that around 90 % or more accuracy of fish identification can be done(due to high end GPU) an once the Artificial intelligence algorithm are trained and it can be loaded into agilex (ai based) FPGAs , which will surely increase the performance and accuracy.

Intel is a perfect Solution provider for this.

World accredited HPC/Server grade processors we have.

Industry accredited Machine Vision, Software Eco-system we have.

 

 

 

References:

[1] Ángel J. Rico-Díaz  Juan R. Rabuñal Marcos Gestal Omar A. Mures and Jerónimo Puertas -An Application of Fish Detection Based on Eye Search
with Artificial Vision and Artificial Neural Networks
, MDPI , 2020 PP 14-15

[2] Kristian Muri Knausgard ˚ 1 · Arne Wiklund2 · Tonje Knutsen Sørdalen3,4 · Kim Tallaksen Halvorsen5 Alf Ring Kleiven3 · Lei Jiao2 · Morten Goodwin2

Temperate fish detection and classification: a deep learning based approach , Springer 2021.


 

 

 
 
While fishing , unwanted fish which is sometimes estimated to be as high as 30% [1]

Poor enforcement, partly due to lack of economical surveillance tools, combined with lack of science-based management, and inadequate global
and local governance reduce global fisheries production by $83 billion annually.
  [2]

 

 

 

 

 

References:

[1] Ángel J. Rico-Díaz  Juan R. Rabuñal Marcos Gestal Omar A. Mures and Jerónimo Puertas -An Application of Fish Detection Based on Eye Search
with Artificial Vision and Artificial Neural Networks
, MDPI , 2020 PP 14-15

[2] Kristian Muri Knausgard ˚ 1 · Arne Wiklund2 · Tonje Knutsen Sørdalen3,4 · Kim Tallaksen Halvorsen5 Alf Ring Kleiven3 · Lei Jiao2 · Morten Goodwin2

Temperate fish detection and classification: a deep learning based approach , Springer 2021.

4. Design Introduction

Tier-1 is our first and foremost eye to address the market.

Tier-1 is 400,00 fishing vessels which are larger in capacity of catchment.

Tier-2 is 1.3 million vessels which are mechanical and larger

Tier-3 is 4 million vessels - which are smaller and unorganized and not in our initial radar.

5. Functional description and implementation

Our venture is coming up with the cutting edge End-to-End product which can help the marine species and over a 5-10 years course wild capture would be rejuvenated naturally with the ultimate solution what we offer with the existing Hardware/Software but integrating and applying it for a unique way.

 

 

Blind Fishing and overfishing has made the marine resources / wild capture as no longer a bottom less fishing.

This overfishing put a trouble to 1/3 of world population especially the under-developed and developing countries who rely ocean as their cheap protein.

30% of natural CO2 removal in this earth will be questioned.

200 million people who are relying on marine ecosystem for generation by generation - will loose their economic freedom and relocate to survival.

Quality/Quantity/Size of catchment from 1920 to 2020 -> has completed changed. Many of the species family in extinct or endangered list. Size has reduced from mega to small size.

By catching the allowable size, fish in abundant and throw-back the fishes which as endangered/extinct list / carrying eggs / small in size(frys) / small in weight.

So we can make sure the natural marine food resource is not damaged and keeping the natural beauty maintained with its original charisma.

 

 

Use the HD Cameras to capture the snap and feed forward to Movidius Vision products / OpenVINO to process eatable/non-eatable

Use the Depth sensor to identify catchable/non-catchable based on different variants of fishes

Use 2D-3D Ultrasound scanners to identify the catchable and eatable fish is carrying eggs or not.

All these data will be feed-forwarded to the Robot-ARM and identified objects from above 3 phases will be thrown back to Ocean.

All these has to be done in fraction of second and at max in terms of minutes to the whole lot of catchment as this process playing with the live captures. So we have to use the Xeon-D based Edge Computing for efficient processing.

6. Performance metrics, performance to expectation

 

Solution space will vary from US $25K to $50K.

 

Phasewise costing :

Phase0: No PC / No Edge.AI / No IP Cams (Ideation)

Feed offline datas & train the system and create your trained database

Phase1: Incubation

Edge.AI --> $2000

Internal GPU (TGL / ADL --> GPU as part of internal)-

/Discrete GPU (Nvidia, Intel DG1/DG2)

IP POE Cams --> $300 - $600 (1 to start)

Phase2: performing / scaling

FPGA --> AI activities --> GPU EU (Execution Units) --> Meta Data --> GPU-

Here fpga agilex  costing - $6,044.95

 

Total around $ 25,000  per device.

 

ROI:

Currently at the fishing vessel, all process is manual of sorting, post interacting with few vessel owner,

It can increase there profits by $20 M per year  and  it increase 10 % YOY.   (considering teir1 ,2, 3, 4 customers)

 

 

Since there are no competitors - can not measure how we are better in comparative.

This is a unique and niche opportunity to penetrate and prove Intel leadership and explore new options to sell more products.

There is a high chance to apply lots of patents as part of integration and end-to-end solution.

Some of real-time processing can be resulted to uplift the IOTG thought process on market and TAM.

 

 

Tier-1 is our first and foremost eye to address the market.

Tier-1 is 400,00 fishing vessels which are larger in capacity of catchment.

Tier-2 is 1.3 million vessels which are mechanical and larger

Tier-3 is 4 million vessels - which are smaller and unorganized and not in our initial radar.

 

 

More over we can convert this solution and extend it to

1. AIS - Alien Invasive Species in salt/fresh water spotting and controlling it.

2. Waste segregation. Around the world this is done in mainly manually or not so organized way. Major issue facing by 2/3 of world countries.

3. Man-less, automated solar operating Plastic and floating waster aggregator in salt/fresh water. ( which is a big head-ache to developing and under-developed countries)

So this is an ever-growing and limit-less solution where we can apply and extend the object spot/train/detect.

7. Sustainability results, resource savings achieved

already submitted block diagram

8. Conclusion

0 Comments



Please login to post a comment.