Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable
Thesis (MEng) -- Stellenbosch University, 2022.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA |
| Published: |
2022
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613755050819584 |
|---|---|
| access_status_str | Open Access |
| author | Van der Westhuizen, Markus |
| author2 | Hummel, Vera |
| author_browse | Hummel, Vera Van der Westhuizen, Markus |
| author_facet | Hummel, Vera Van der Westhuizen, Markus |
| author_sort | Van der Westhuizen, Markus |
| collection | Thesis |
| description | Thesis (MEng) -- Stellenbosch University, 2022. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/125957 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:41:10.692Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2022 |
| publishDateRange | 2022 |
| publishDateSort | 2022 |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/125957 Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable Van der Westhuizen, Markus Hummel, Vera Von Leipzig, Konrad Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. Agricultural innovations Machine learning Augmented reality Fruit -- Breeding Agriculture -- Economic aspects UCTD Thesis (MEng) -- Stellenbosch University, 2022. ENGLISH ABSTRACT: Advancements in Industry 4.0 technologies are leading to the digitalisation and automation of the agricultural sector. The risk of automation is that it could lead to the marginalisation of small and medium sized farms as well as an increase in unemployment. Therefore, Augmented Reality, with advanced visual capabilities, was studied as it has the potential to aid employees in the agricultural sector instead of displacing them. Within the fruit industry, avocado farms were selected as the focal point to test Augmented Reality. This was due to avocado farms providing significant improvement opportunities, with 50% of fruit being wasted, and the potential to increase the economic activity of very poor regions globally, as the climate to grow avocados are in these regions. The productivity and quality improvements were shown to be the most advantageous for farmers who still pack avocados on their farm, which is currently a manual process which could benefit from the collaboration between human and technology. To test the possible productivity and quality improvements to avocado fruit classification process a prototype was developed. This prototype utilised Augmented Reality and Machine Learning technologies to assist a packer when classifying avocados according to size and grade, to improve the accuracy and speed of avocado classification. The practical implementation of Augmented Reality and Machine Learning technologies were done using the HoloLens 1 and Microsoft Azure respectively. Post development the prototype was tested, and the results showed a significant increase in packing quality as the accuracy increased from 73.3% when grading and 58.5% when sizing avocados to 83.0% and 73.3% respectively. To test the productivity improvement of the prototype both the packing speeds of a trained and untrained packer were evaluated with and without the prototype. The productivity improvement (measured in the packing speed increase) for the trained and untrained packers were 29.87% and 54.88% respectively when utilising the HoloLens 1. The results showed that both quality and productivity improvements can be made when Augmented Reality is implemented on avocado farms. Having proven that both the quality and productivity of the avocado classification process have been improved it was then necessary to test the possible financial benefit that can be generated which would make retaining labour economically viable. The economic results show that for small and medium sized farmers who either pack their own fruit on the farm or utilise a packing facility there is a significant financial benefit when utilising the prototype. It is found that for small and medium sized farmers who currently pack their own fruit, who are also the mostly likely adopters of and would benefit the most from the prototype, the financial benefit is R438 426.99 and R2 192 134.96 respectively per annum. Therefore, the prototype, utilising Augmented Reality and Machine Learning technologies, was able to increase the quality and productivity of the fruit classification process, resulting in significant economic benefit, justifying the retention of labour on avocado farms, and potentially the entire agricultural sector. AFRIKAANS OPSOMMING: Vooruitgang in Industry 4.0-tegnologieë lei tot die digitalisering en outomatisering van die landbousektor. Die risiko van outomatisering is dat dit kan lei tot die marginalisering van klein en medium boere sowel as 'n toename in werkloosheid. Daarom was Verhoogde Werklikheid, met gevorderde visuele vermoëns, bestudeer aangesien dit die potensiaal het om werknemers in die landbousektor te help in plaas daarvan om hulle te verplaas. Binne die vrugtebedryf was avokadoplase gekies as die fokuspunt om Verhoogde Werklikheid te toets. Dit was te danke aan avokadoplase wat verbeteringsgeleenthede bied, met 50% van vrugte wat vermors word, en die potensiaal om die ekonomiese aktiwiteit van baie arm streke wêreldwyd te verhoog, aangesien die klimaat om avokadopere te kweek in hierdie streke is. Die produktiwiteit en gehalteverbeterings het getoon dat dit die voordeligste is vir boere wat nog avokadopere op hul plaas verpak, wat tans 'n handmatige proses is wat voordeel kan trek uit die samewerking tussen mens en tegnologie. Om die moontlike produktiwiteit en gehalteverbeterings in die klassifikasieproses van avokadopere te toets, was 'n prototipe ontwikkel. Hierdie prototipe het Verhoogde Werklikheid en Masjienleer-tegnologie gebruik om 'n pakker te help om avokadopere volgens grootte en graad te klassifiseer, om die akkuraatheid en spoed van avokadopereklassifikasie te verbeter. Die praktiese implementering van Verhoogde Werklikheid en Masjienleer-tegnologie was met onderskeidelik die HoloLens 1 en Microsoft Azure geïmplementeer. Na ontwikkeling was die prototipe getoets, en die resultate het 'n beduidende toename in verpakkingsgehalte getoon, aangesien die akkuraatheid van 73.3% waneer graad gradering en 58.5% waneer grootte gradering van avokadopereklassifikasie tot 83.0% en 73.3% onderskeidelik gestyg het. Om die produktiwiteitsverbetering van die prototipe te toets, is beide die pakspoed van 'n opgeleide en onopgeleide pakker geëvalueer met en sonder die prototipe. Die produktiwiteitsverbetering (gemeet in die pakspoedverhoging) vir die opgeleide en onopgeleide pakkers was onderskeidelik 29.87% en 54.88% met die gebruik van die HoloLens 1. Die resultate het getoon dat beide 'n beduidende gehalte- en produktiwiteitsverbetering gemaak kan word wanneer Verhoogde Werklikheid geïmplementeer word op avokadoplase. Nadat dit bewys was dat beide die gehalte en produktiwiteit van avokadopereklassifikasie verbeter kan word, was dit dan noodsaaklik om die moontlike finansiële voordeel te toets wat gegenereer kan word wat die behoud van arbeid ekonomies lewensvatbaar sou maak. Die ekonomiese resultate toon dat vir klein- en mediumgrootte boere wat óf hul eie vrugte op die plaas pak óf 'n verpakkingsfasiliteit gebruik maak, daar 'n aansienlike finansiële voordeel is wanneer die prototipe gebruik word. Daar was gevind dat vir klein- en mediumgrootte boere wat tans hul eie vrugte verpak, wat ook die mees waarskynlike aannemers van die prototipe is en die meeste daarby sal baat, die finansiële voordeel R438 426.99 en R2 192 134.96 onderskeidelik per jaar kan wees. Daarom was die prototipe wat gebruik gemaak het van Verhoogde Werklikheid en Masjienleer-tegnologie in staat om die gehalte en produktiwiteit van die vrugteklassifikasieproses te verhoog, wat beduidende ekonomiese voordeel tot gevolg gehad het wat die behoud van arbeid op avokadoplase, en moontlik die hele landbousektor, regverdig. Masters 2022-11-14T15:39:36Z 2023-01-16T12:43:00Z 2022-11-14T15:39:36Z 2023-01-16T12:43:00Z 2022-12 Thesis http://hdl.handle.net/10019.1/125957 en_ZA xvii, 120 pages : illustrations application/pdf |
| spellingShingle | Agricultural innovations Machine learning Augmented reality Fruit -- Breeding Agriculture -- Economic aspects UCTD Van der Westhuizen, Markus Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title | Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title_full | Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title_fullStr | Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title_full_unstemmed | Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title_short | Using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| title_sort | using visual technology to increase the productivity and quality of fruit classification in order to make retaining labour economically viable |
| topic | Agricultural innovations Machine learning Augmented reality Fruit -- Breeding Agriculture -- Economic aspects UCTD |
| url | http://hdl.handle.net/10019.1/125957 |
| work_keys_str_mv | AT vanderwesthuizenmarkus usingvisualtechnologytoincreasetheproductivityandqualityoffruitclassificationinordertomakeretaininglaboureconomicallyviable |