AI-IoT enable Jute fibre grading system

Background/Problem Description:

Jute is one of the important commercial crops in India, primarily cultivated for its fiber. India and Bangladesh are the leading producers, accounting for 95% of the world’s jute production. India alone contributes about 50% of the global jute output. In India, jute is predominantly cultivated in the eastern regions, specifically in West Bengal, Bihar, Odisha, Assam, Tripura, Meghalaya, Andhra Pradesh, and Telangana. Among these, West Bengal, Assam, and Bihar collectively produce nearly 99% of India’s jute. West Bengal alone represents 80.2% of the total jute acreage and 82.3% of the country’s production. Further, the jute sector supports the livelihood of around 4 million farm families in India and provides direct employment to 3,70,000 workers in organized mills and the diversified sector.

Due to various factors such as plant variety, age, soil condition, retting process, extraction method, and drying techniques, the quality of jute is not uniform across India. This variability necessitates a classification/grading system to ensure consistent quality standards. Therefore, grading the fiber is crucial for commercial purposes, as the Minimum Support Price (MSP) of jute depends on its grade. The Bureau of Indian Standards has recommended a grading system for procurement. In India, the prevalent grading method is the "hand and eye" system, where expert graders visually assess the fiber. Despite the availability of digital instruments for jute grading, this traditional method persists, leading to biases and non-scientific evaluations that often deny farmers a fair price for their produce. Testing with available digital grading instruments is time-consuming, as sample preparation takes a significant amount of time. To expedite the grading of jute fiber, ICAR-NINFET in Kolkata developed an integrated grading system for natural fiber employing traditional image technique and mechanical setup. Although this system can grade the fiber, one of the five parameters, root content, still needs to be manually measured and entered into the program. Additionally, the varied performance characteristics of aforesaid instrument. Advancement of artificial intelligence and IoT has made several complex tasks easy especially in data processing and data transfer. It is thus there has been a need to develop Artificial Intelligence-IoT enabled jute fibre grading system to measure the individual parameters of jute fibre accurately and determine the final grade of fibre.

Technology Details:

Grading jute fiber is essential in determining the quality and price of the final product. In India, the prevailing method of grading is the hand and eye approach, which is subjective and lacks scientific basis. However, the development of digital instruments offers a more objective and scientific assessment of jute fiber. Testing with available digital grading instruments is time-consuming, as sample preparation takes a significant amount of time. To expedite the grading of jute fiber, ICAR-NINFET in Kolkata developed an integrated grading system for natural fiber employing traditional image technique and mechanical setup. Although this system can grade the fiber, one of the five parameters, root content, still needs to be manually measured and entered into the program. Additionally, the varied performance characteristics of aforesaid instrument. The developed Artificial Intelligence-IoT enabled jute fibre grading system expedites the jute grading, eliminate manual intervention and accurate.

The Artificial Intelligence-IoT enabled jute fibre grading system consists of several components: a feeding tray, a conveyor, an image capturing unit, a colour measurement setup, and a bundle strength measurement unit. The feeding tray, made of aluminium, is used to load the fibre into the conveying system. The conveyor moves the fibre at a fixed distance under the image capturing unit, which uses a 12.3 Megapixel Raspberry Pi camera with a rolling shutter, captures images as the fibre passes under it. These images are automatically sent to a cloud server for processing. Three of the five parameters are measured using these captured images, which are analyzed through a computer vision-based contour detection model. The colour measurement setup includes four BPW 34 photodiode sensors and 5 mm light-emitting diodes. The bundle strength measurement unit features a fibre breaking unit with a limit switch, motors, an S-type load cell with an amplifier, and an Arduino microcontroller. This unit employs three motors: one to exert a constant loading rate through the jaw base, and the other two to facilitate the jaw’s to-and-fro motion. After measuring all parameters, the mobile app "Jute Grader" displays the test results. Once confirmed, the results are sent to and stored on the cloud server. A distinguishing feature of this jute grading instrument is its ability to measure all parameters of the jute within five minutes, providing both individual values and the final grade of the jute fibre.