Search the GenAI Directory

Fine tuning a model for crop advice needs training examples that someone has checked rather than raw scraped text. This is a downloadable collection of roughly seventeen thousand crop and soil instruction and answer records that went through a review step.

Checking pack copy, allergen wording and nutrition panels against the rules for each market eats reviewer time and risks a recall. This is a hosted review tool that reads your packaging files and scores them against rules you write in plain language.

Dietary studies and food logging apps need images where each food item is located, not just named. This is a food photograph dataset with 82 categories and at least one marked region per picture.

Driving a prototype robot down real maize rows for testing is only possible a few weeks a year. This is a package that builds randomised virtual maize fields inside a robot simulator, so navigation code can be tried repeatedly.

Working a grain cargo means running the same negotiation with many counterparties at once and then writing it all up. This is a hosted assistant that handles those chat negotiations, combines cargoes and drafts contracts and trade confirmations. The publisher describes it in the context of trading in chat.

Setting up a fish or shellfish farm means choosing tanks, pumps and water handling before any stocking. This is a preconfigured chat assistant, opened through a shared link, focused on aquaculture system design, equipment and husbandry.

Livestock projects often need a model already tuned to animal imagery rather than general photos. These are openly licensed downloadable model weights, with code, from a Hugging Face repository specialised for livestock.

Hunting down weed imagery across scattered research projects wastes time. This is a web portal collecting annotated weed photo sets from various field studies, including well known open collections, with downloads in one place.

Telling weeds from crops in drone or plane imagery is the basis for spot treatment. This is an aerial image dataset of weeds and crops, listed in an open catalogue and reachable through a web page.

You need many more weed photos than your field trials produced before a recognition model has enough to learn from. This is a generator with downloadable weights and notebooks that produces artificial weed images on your machine or in the cloud.

Anyone training software to spot weeds needs annotated field photographs, which are not easy to collect yourself. This is a publicly shared image dataset of weeds, usable for training, testing or as background material.

Farmers watching their wells and deciding when to irrigate rarely have an expert on call. This is a chat helper built into a well monitoring phone app that answers questions in English or Arabic about water use and irrigation choices. The publisher describes it in the context of in My FarmWell v1.3.0.

Prototyping retail vision work is awkward before you have access to a picture library. This is a small sample of grocery shopper photographs with accompanying details, drawn from a larger picture collection.

Writing up spraying and fieldwork records at the end of the day rarely actually happens. This is a WhatsApp assistant that turns typed or spoken messages into structured field records in your farm account and sends updates back.

Spotting what is wrong with a leaf usually means waiting for someone who knows. This is a downloadable model that looks at a plant photo and describes possible leaf disease in words, with code to run it.

Catching a pest in maize early is hard if you are not sure what you are looking at. This is a small hosted demo where you upload a photo and a model suggests which maize pest it shows, with the files published.

Food questions often come with a picture, whether it is a dish or a label. This is an open code repository, with a programming interface, for an assistant that works with food related text and images.

Testing a plant DNA model across several kinds of questions means hunting down a different dataset for each one. This is a bundle of more than ten downloadable plant sequence datasets covering things like gene switches, chromatin and long non coding RNA.

Growers who want farming advice in Arabic often cannot reach an adviser when the question comes up. This is a chat feature inside an Arabic farm advice app that answers questions in everyday language across different farming topics.

Growing questions depend heavily on your region and the variety you have planted. This is a paid feature inside a Japanese farm business platform offering round the clock cultivation advice that takes area and variety into account. The publisher describes it in the context of AgriweB premium service.

Agricultural knowledge in Taiwan sits in thousands of reports and theses nobody has time to read. This is a hosted question answering service, built on a Traditional Chinese language model, that searches that library and cites its sources.