Why aren’t farmers using new tech?

“Agtech, short for agriculture technology, is a growing industry that’s using data tools and software to help farmers improve yields and use fewer resources.

“With population growth increasing the global demand for food and climate change hurting crop yields, a swift adoption of agtech may be needed now more than ever. Yet, farmers are hesitant about embracing these new technologies.

“What’s in the way of farmers quickly adopting agtech, and how can the industry get more farmers on board?

“Marketplace” host Kai Ryssdal talked to reporter Belle Lin from the Wall Street Journal about her recent article on why so few farmers are using agtech. Below is an edited transcript of their conversation.”

Read the full Marketplace interview here…

Also, America’s Farmers Are Bogged Down by Data

“Uptake of agtech tools has been tepid, and even many farmers who do use them struggle with the software and a flood of data from their farms” from the Wall Street Journal, read the full here…


Evaluation of Computer Vision Systems and Applications to Estimate Trunk Cross-Sectional Area, Flower Cluster Number, Thinning Efficacy and Yield of Apple

by Luis Gonzalez Nieto 1, Anna Wallis 2, Jon Clements 3, Mario Miranda Sazo 4, Craig Kahlke 4, Thomas M. Kon 5 and Terence L. Robinson 1

1 Horticulture Section, School of Integrative Plant Science, Cornell University, Geneva, NY 14456, USA
2 Extension Section, Michigan State University, Grand Rapids, MI 49503, USA
3 UMass Extension Section, University of Massachusetts, Belchertown, MA 01007, USA
4 Cornell Cooperative Extension, Lake Ontario Fruit Program, Albion, NY 14411, USA
5 Mountain Horticultural Crops Research and Extension Center, Department of Horticultural Sciences, North Carolina State University, Mills River, NC 28759, USA

Horticulturae 2023, 9(8), 880;

Precision crop load management of apple requires counting fruiting structures at various times during the year to guide management decisions. The objective of the current study was to evaluate the accuracy of and compare different commercial computer vision systems and computer applications to estimate trunk cross-sectional area (TCSA), flower cluster number, thinning efficacy, and yield estimation. These studies evaluated two companies that offer different vision systems in a series of trials across 23 orchards in four states. Orchard Robotics uses a proprietary camera system, and Pometa (previously Farm Vision) uses a cell phone camera system. The cultivars used in the trials were ‘NY1’, ‘NY2’, ‘Empire’, ‘Granny Smith’, ‘Gala’, ‘Fuji’, and ‘Honeycrisp’. TCSA and flowering were evaluated with the Orchard Robotics camera in full rows. Flowering, fruit set, and yield estimation were evaluated with Pometa. Both systems were compared with manual measurements. Our results showed a positive linear correlation between the TCSA with the Orchard Robotics vision system and manual measurements, but the vision system underestimated the TCSA in comparison with the manual measurements (R2s between 0.5 and 0.79). Read more here…


Autonomous Apple Fruitlet Sizing and Growth Rate Tracking using Computer Vision

Harry Freeman1, Mohamad Qadri1, Abhisesh Silwal1, Paul O’Connor2, Zachary Rubinstein1, Daniel Cooley2, and George Kantor1

1Carnegie Mellon University Robotics Institute, PA, USA

2University of Massachusetts Amherst Stockbridge School of Agriculture, MA, USA

Abstract – Measuring growth rates of apple fruitlets is important because it allows apple growers to determine when to apply
chemical thinners to their crops to optimize yield. The current
practice of obtaining growth rates involves using calipers to
record sizes of fruitlets across multiple days. Due to the number
of fruitlets needed to be sized, this method is laborious, timeconsuming, and prone to human error. In this paper, we present
a computer vision approach to measure the sizes and growth
rates of apple fruitlets. With images collected by a hand-held
stereo camera, our system detects, segments, and fits ellipses to
fruitlets to measure their diameters. To measure growth rates,
we utilize an Attentional Graph Neural Network to associate
fruitlets across different days. We provide quantitative results on
data collected in an apple orchard, and demonstrate that our
system is able to predict abscise rates within 3% of the current
method with a 7 times improvement in speed, while requiring
significantly less manual effort. Moreover, we provide results on
images captured by a robotic system in the field, and discuss the
next steps to make the process fully autonomous.

Read more here…


Publication:  Digital Technologies for Precision Apple Crop Load Management (PACMAN) Part I: Experiences with Tools for Predicting Fruit Set Based on the Fruit Growth Rate Model 

 Anna Wallis, Jon Clements, Mario Miranda Sazo, Craig Kahlke, Karen Lewis, Tom Kon, Luis Gonzalez, Yu Jiang and Terence Robinson 

Reprinted from Fruit Quarterly, Volume 31, Number 1, Spring 2023

Decades of work have demonstrated that PACMAN (Precision Apple Crop load MANagement) is an extremely effective method for successfully managing crop load. Effective crop load management has a direct effect on yield, quality, size, and return bloom, and ultimately an orchard’s profitability. The process involves three management practices: 1) pruning, 2) chemical thinning, and 3) hand thinning, which have been described in detail in previous articles (Robinson et al., 2014a,b). We are continuing to refine recommendations for PACMAN, on a regional basis, as part of a 4-year national project, funded by the USDA-NIFA SCRI. This article is a follow-up to our previous article summarizing earlier work on this project (Robinson et al., 2022). 

Full article below:


It’s a wrap…

The last of the PACMAN show & tell’s are posted on the WSU CAHNRS YouTube channel here: These now include updates from the PACMAN engineering team, including:

14 – 3-D Imaging and Digital Twin for Specialty Crops (Yu Jiang, Cornell University)

15 – Crop Load Adjustment Based on Early Flower Detection (Long He, Rashmi Sahu, and Paul Heinemann, Penn State)

16 – Improving Apple Harvest with the Latest in AI Yield Estimation for Specialty Crops (Dana Choi, University of Florida)

17 – Moog’s Autononmous Journey: “Agriculture is the Lynch Pin” (Chris Layer and Tom Fischer, Moog, Inc.)




 Fruit set prediction models aim to produce timely estimates of fruitlet abscission after thinner applications to guide precision crop load management. The time to generate a prediction after an application is important to facilitate grower decisions to re-apply thinners while they are still efficacious, avoiding expensive hand thinning operations. The fruitlet growth rate (FGR) model, developed by Dr. Duane Greene, is a powerful tool that can accurately predict the percentage of fruitlets that will set in an orchard. Although a Excel data template and App are available to run the FGR model via computer and smartphone, respectively, adoption has been limited by the measurement-intensive procedure. A new approach, termed the ‘Fruitlet Size Distribution (FSD) Model’, described herein, was developed to produce predictions of apple fruit set comparable to the FGR model but achievable with less time investment. The principle underlying both models is the same: the relative growth rate or size of a fruitlet is compared to the most rapidly growing or largest fruitlet within the sample date to determine if it will abscise. Most predictions can be made within 8 days from thinner applications, though the duration of time depends on climatic, biological and horticultural factors. To optimize the FSD model we suggest beginning the model three days after the average fruitlet diameter of the orchard is 6 mm. Thus, the model partners well with thinning applications between bloom and 6 mm. For example, if, a prediction can be achieved by 8 days, assuming an average growth of ~0.8 mm per day, then fruitlets will be ~ 12 mm if another application is needed; 12 mm fruitlets are very sensitive to many thinning chemistries.

Complete directions and Excel file linkg for using the FSD Model can be downloaded below…

XLSM file for calculating FSD model here. Please read HOW-TO-GUIDE above. Note the XLSM file will need to be downloaded and run locally on your computer. Macros have to be enabled.


PACMAN Briefings – Session IV

Beyond Commercialization – Engineering Research for Crop Load Management

March 9 @ 9:00 am – 11:00 am (Pacific Time)

This session provides updates on engineering-related research that looks to the future for apple crop load management. See what is being developed for further sensing and automation to assist with orchard production. Presentations are provided by researchers at Cornell University, Penn State, University of Florida, and Moog, Inc. These will be in a short 10-minute format, with time for questions and discussions at the end.


Karen Lewis, Agriculture and Natural Resources Program Unit Director, Tree Fruit Extension Specialist, Washington State University

3D Imaging and Digital Twin for Specialty Crops

Yu Jiang, Assistant Research Professor, Agritech, Cornell University

Crop Load Adjustment Based on Early Flower Bud Detection

Long He, Assistant Professor of Agricultural and Biological Engineering, Rashmi Sahu, Graduate Student Paul Heinemann, Professor of Agricultural and Biological Engineering, Penn State

Improving Apple Harvest with the Latest in AI Yield Estimation for Specialty Crops

Dana Choi, Assistant Professor of Agricultural Engineering, University of Florida

Moog’s Autonomous Journey: “Agriculture Is the Lynch Pin”

Chris Layer, Mgr. Technology & Advanced Pursuits, and Tom Fischer – Moog Inc.

Advance registration required…

WSU Tree Fruit Event Site


Mask R-CNN based apple flower detection and king flower identification for precision pollination

“Identification of individual king flowers within flower clusters is a critical step for developing a robotic apple pollination system. Typically, each cluster has five to six individual flowers, and the king flower can be occluded by the lateral flowers because of their central position in a flower cluster. King flowers share identical features (e.g., color, shape, and size) with other flowers. Apple flower clusters open sequentially from the king flower to the lateral flowers in the time of anthesis, which presents an opportunity for selective pollination. Therefore, it is critical to monitor the flower blooming stage for accurately determining the pollination targets and timing…”

Xinyang Mu, Long He, Paul Heinemann, James Schupp, and Manoj Karkee

Read more at Science Direct


PACMAN Briefings Chat Transcript 3

January 26, 2023 Chat Transcript

Recording(s) here

Today’s Guests: Oli Hilbourne (Outfield); Jenny Lemieux (Vivid Machines); Karina Lau (AgerPix); Steve Mantle (Innov8); Mauricio Guerra Funes (Cornell University)

00:45:26 GK: Any use of drones for pest scouting?
00:45:53 BA: How big does fruiot have to be to detect in mm
00:46:35 BA: How small can you define a zone to 2′?
00:46:35 MJM: Is the drone useful for all training systems?
00:46:58 JC: Can the drones fly approx. 2m off the ground between rows to take sideview photos of the tree (for flower and fruitlet counts)
00:47:25 LS: How does your drone technology handle adverse weather conditions such as high winds and heavy rain?
00:49:12 Oli Hilbourne | Outfield: Hi Bruce. The fruit needs to be 40mm or about 1 inch in diameter to be picked up with the Outfield system. We can count from from 40mm right up to harvest.
00:50:46 Oli Hilbourne | Outfield: Hi Md Jebu. The drone works on trellis wall and super spindle growing systems. This year we will be demonstrating the system on V trellis and V4 systems.
00:53:11 Oli Hilbourne | Outfield: Hi John. We typically fly around 10-12m above the ground to survey the crops. We are always looking at higher resolution cameras, as as those become more common hopefully we will be able to pick up fruit earlier and earlier, right up to fruitlet counting. At the moment its only blossom cluster mapping and fruit counting from 40mm diameter and larger.
00:55:05 Oli Hilbourne | Outfield: Hi Lina. The drone works well in all lighting conditions, and being based in the UK we have plenty of examples of rainy conditions to train on! The drones don’t fly in the rain, and you shouldn’t fly in winds of 20 mph or more.
00:55:34 LS: Reacted to “Hi Lina. The drone w…” with 👍
00:56:49 JC: Jenny – what is the smallest fruit size your camera can measure?
00:57:17 RM: Do you have grower in Quebec
00:58:51 MJM: Hi Jenny, how about the price of the camera?
00:59:13 RF: Jenny- Have you started work on canopy density
01:03:39 Jenny Lemieux | Vivid Machines: Hi MJM – we lease the sensor for $5k/yr, and charge $60/acre for the predictions and services.
01:04:44 Jenny Lemieux | Vivid Machines: Hi Rod – we haven’t built a model yet for canopy density, mostly because we need to determine how to define it (per block, or variety, per farm.. ). but we have the data needed and it’s on the list! We just need help with thinking through what to present.
01:08:15 SB: How many dendrometers have been deployed in one orchard to get accurate growth curves? Or is there a recommendation of number of units per acre?
01:09:54 SB: Replying to “How many dendrometer…” answer: 6 units/hectare
01:11:10 LCF: Replying to “How many dendrometer…” Always depends on the species, the age, etc. One antenna covers around 50 Has.
01:24:43 RM: hi Jenny I assume that we must drive each sides of the row to have accurate data. What is the max speed of the 4 Wheel to obtain good results
01:26:04 IL: Hi Robert, yes driving both sides of the row is key to obtain a holistic view of your block. The Green Atlas cartographer can be driven at speeds of 25mph
01:26:29 AAK: Hi
01:28:30 Jenny Lemieux | Vivid Machines: hi Jenny I assume that we must drive each sides of the row to have accurate data. What is the max speed of the 4 Wheel to obtain good results Hi RM – no, just one side. One of the algorithms in our pipeline deals with occlusion so that we can predict from the one side.
Max speed is 10 km/hr, though on 3 and 4 leaf trees, a little slower, 5-7km is currently working better (though I expect we’ll solve that before April)
01:35:38 HGT: Does any of the speakers do the imaging/dat collection by aerial application?
01:38:20 Steve Mantle: HGT – partnerships roadmapped for innov8 – satellite (planet), flyover, drone (perhaps with some of the PACMAN presenters). Our belief is none of these systems are a one size fits all; and each of these methodologies (including handheld data capture) have a place to complement depending on extent of variability and need to scale.
01:39:03 Oli Hilbourne | Outfield: Hi HGT. Outfield uses drones to gather imagery data, we use that to create blossom loading maps, fruit counts and yield estimates. We use off the shelf consumer drones that growers can fly on demand. I was the first speaker on today’s call, I think there will be a video available later but I’m happy to follow up directly if you like.
01:40:42 HGT: my apology: clarification I am thinking fixed wing aircraft, helicopter or satellite. , not drones.
01:41:12 Oli Hilbourne | Outfield: Ah, no problem!
01:43:08 SB: Accuracy
01:43:22 AKL: On point
01:43:35 RK: total cost of ownership needs to be added
01:43:41 JMR: Results delivered in a timely manner
01:43:57 SB: Durability (could be considered ease of use)
01:44:00 SB: I think how the various technologies integrate with the others (new or existing ones) is important
01:44:26 JL: Net profit results
01:44:40 KB: what info we can gather – precision thinning vs final yields
01:45:47 SB: everybody is working on adding new tools, but each living in its own space, with no common/shared way of communicating with other tools or existing dataset
01:46:49 SB: yes
01:46:56 Steve Mantle: SB – PRECISELY. That’s our focus – bringing together data siloes (selectively, without it being too crazy!)
01:47:24 HGT: cost benefit analysis
01:47:45 SB: Units of cost is also important like is it per user, per acre, per year, etc.
01:47:46 Karen Lewis: Reacted to “everybody is working…” with 👍🏻
01:48:31 Steve Mantle: API availability for data flow is critical.
01:48:32 SB: Steve, I have a feeling no one company can solve this issue, but there should be an effort among tech companies to define a common ground, some standards
01:48:32 HGT: again: cost benefit analysis
01:48:55 HGT: it is all about DATA and their benefit
01:49:15 JL: Different for each
01:49:15 SB: mmmmm….maybe 100-500
01:49:17 Steve Mantle: SB – 100% agree. Getting us all oaring the right direction and aligning on common items like task maps are key.
01:50:20 RK: based on only 1 factor, is not too helpful for me.I want to see the overall picture. Chemical savings, environmental benefits…..
01:50:39 HGT: yes
01:50:44 SB: yes
01:50:54 JL: Yes
01:50:56 MJM: yes
01:50:59 AAK Dyes
01:51:06 ADR: turn on and receive data
01:51:14 DVW: Yes
01:51:26 JMR: Reacted to “turn on and receive …” with 👍
01:51:28 RK: Turn on and receive data if I can massage raw data for my use.
01:52:56 JL: Thanks for these sessions. Very enlightening!
01:53:49 Steve Mantle: Hats off to the organizing team for this series, and the tons of work you’re doing behind the scenes.
01:53:53 Craig Kahlke: Past PACMAN Meetups (Jan 12., 19, and today’s in the coming days):
01:54:17 Oli Hilbourne | Outfield: Thank you very much to the organising team for putting this all together, and thank you for the invite to speak today!
01:54:37 Craig Kahlke: W NY Fruit Conference – “State of the Industry” Feb 27-8 Rochester, NY
01:55:04 AKL: thank you PACMAN Extension team
01:55:15 Anna Wallis: Link to the PACMAN website for additional project updates:
01:55:19 Jenny Lemieux | Vivid Machines: Thanks everyone!
01:55:26 Karen Lewis:
01:55:27 GS: thx!!
01:55:29 SB: will you have web connection for the conference?
01:55:39 ADR: thanks!
01:55:45 Steve Mantle: Yes – would love to engage more providers in smart orchard 🙂
01:55:58 RK: thank you all
01:56:03 MG: Thank you !
01:56:49 SB: is there a newsletter for the PACMAN project?
01:56:56 Steve Mantle: Where do we subscribe to the quarterly?
01:57:13 Karen Lewis: Replying to “is there a newslette…” Pacman website – no newsletter
01:57:33 Jon Clements: No newsletter, website has blog/post updates
01:57:51 Anna Wallis:
01:57:51 SB: Reacted to “No newsletter, websi…” with 👍🏻
01:58:03 Steve Mantle: Reacted to “…” with 😁
01:58:15 SB: thanks!
01:59:26 Steve Mantle: Turning around the links of recordings in 24 hours is awesome; thank you!
01:59:49 AAK Thankfully


PACMAN Briefings Chat Transcript 2

January 19, 2023 Chat Transcript

Recording(s) here

Today’s Guests: Dave Brown and Patrick Plonski (Pometa); Chris Hall and Matt King (FruitScout); Ross Kranz and Tim Cromwell (LaGasse, Aurea, Munckhof); Charlie Wu (Orchard Robotics)

00:31:00 Anna Wallis:
00:31:36 Anna Wallis:
00:42:09 Craig Kahlke: Please use the chatbox for questions folks.
00:42:48 GP: So, you’re confident you’re not double counting fruits like opposite row on “V” trellis with 3D image.
00:43:40 Patrick Plonski: Hi GP, Yes we are quite confident in this. We can reconstruct which side of the row any individual fruit is growing on. E.g. which arm of the V trellis.
00:44:23 JCL: Are the light conditions in the orchard import for taking video and achieving accurate fruit size measurements.
00:45:42 JC: Could you briefly touch on bin scanning? Like to you use a iPhone with video to scan bins or take an image. Does it get size and color data?
00:46:15 Greg Peck: How will fruit number be reported? Per linear length of row, per tree, per TCSA, per acre, etc.?
00:46:17 Put your name please!: Is there a way to apply a treatment to trees that would enhance buds and make identifying buds a doable
00:47:16 Anna Wallis: Anonymous question: Can you disclose the average percentage of occluded fruit?
00:47:49 MK: How many trees/feet do you need to count for ground truthing?
00:48:22 Put your name please!: Is there an auto-weight calculation based on drop estimates (tonnage)? Any accuracy data therefrom?
00:48:54 JCL Does your output include fruit per tree or fruit per linear foot?
00:48:57 Jon Clements: Greg Peck, Pometa tells me they are going to be focusing on a per tree count going forward, TBD?
00:52:54 Greg Peck: Reacted to “Greg Peck, Pometa te…” with 👍
00:57:00 RB: Can you say if the higher the density of the canopy, the more you overestimate the crop?
01:04:53 Patrick Plonski: Hi JC, Regarding bin scanning, Pometa will be implementing two bin scanning procedures, both using video from the phone. Customers will be able to scan a bin in the field using a phone in much the same manner they can scan a set of trees. We are also testing out overhead scanning so that the packing house can scan the top of the truck as it comes in with a fixed overhead phone.
01:06:09 JC: I’m assuming a iPad Pro would work as well?
01:10:37 BB: How accurate are the trunk diameter measurements with FruitScout, and how robust is it at dealing with interference from ground covers (weeds), root suckers, etc.
01:11:49 Chris Hall: Hi JC. Yes, modern iPad pro cameras would be sufficient, though you may find a phone (iPhone or Android) may be easier to manage just from a size perspective.
01:12:43 Chris Hall: Hi BB, we have found trunk and fruit measurements to be very accurate. Ground truthing done by us and academic and research partners has shown less than 2% variance
01:16:29 SS: For Matt King: HI, could you please repeat how your fruit sizing technology is linked to DAS, I did not get it, thanks
01:19:26 Matt King: Sure. For DAS customers in WA, you can just take sizing pictures with FruitScout, and then we automatically feed that data into DAS and you can see the harvest projections in DAS. For customers outside WA, we integrate with any weather stations, and implement the exact same harvest prediction model. You can see the results in the FruitScout PCLM Control Panel.
01:29:18 Put your name please!: How much cost the root pruner
01:29:26 BB: are they currently selling the VR fertilizer spreaders?
01:34:52 SB: As far as I know there is no automatic integration of FruitScout with DAS. Matt King can you please clarify that?
01:39:26 Matt King: Hey SB. Looks like you are on the emails from last year when we first built the integration with DAS.
01:43:10 BA: Any issues with time of day scan was made
01:45:01 SB2: Can you explain what form the grower receives the data (i.e. excel?)
01:46:03 SS: Hi Charlie, can we use your technology without a vehicle? Like manually? do you have a protocol in place for this options or not possible?
01:47:55 SB: Hey Matt, yes, but I’m still not sure what you mean with integration with DAS. As of today, DAS does not interface directly/automatically with FruitScout, nor vice versa
01:48:31 SS: thanks!
01:51:29 Matt King: SB, I’ll have to beg to differ, as the integration was built, demonstrated last year and exists. But, we can discuss this further offline.
01:51:45 Dave Brown: Thanks Anna!