Plot quality traits at scale in breeding trials to optimize resources and correct yield.

Assessing plot quality in a high-throughput fashion using drone imagery.

In every breeding program, plot quality plays a decisive role in how reliable and comparable the results are. Missing plants, uneven emergence, or damaged zones can all distort yield calculations and compromise data interpretation — yet these issues often go unnoticed until harvest, when it’s too late to adjust.

With drone phenotyping, breeders can now bring objectivity and precision to this essential step. By analyzing high-resolution RGB imagery, Hiphen’s algorithms compute digital traits that automatically detect missing plants, non-uniform stands, and anomalies within each experimental plot. From these indicators, a global quality score is generated for every plot, providing an instant overview of trial integrity.

This digital assessment not only helps breeders optimize harvest logistics and manage trial resources more efficiently, but also supports data correction by adjusting yield results based on the actual harvestable area of each plot.

Consistent and repeatable, this method ensures that no irregularities go unnoticed across locations or seasons. In the end, it’s about giving breeders the confidence that every yield value truly reflects the genetic potential of their varieties — supported by transparent, objective, and quantified field data.

Predict Plant Characteristics to Eliminate F1 crossings most likely to fail.

Predicting trait values of new hybrids way in advance, from the characteristics of the parental lines.

What if you could predict a plant’s performance long before harvest? In modern breeding, this is what’s at stake with phenomic predictions — the ability to foresee key traits early in the trial, or even from parental lines, to streamline the breeding process and shorten time to market.

At Hiphen, we believe this is the true future of phenotyping. By combining digital traits extracted from drone imagery with advanced latent space analysis, we can predict complex agronomic traits such as yield, dry matter content, or flowering date (FLOF) for instance — months before they are traditionally measured.

This predictive capacity enables breeders to eliminate F₁ combinations most likely to fail and focus their resources on the most promising genetic material. When combined with genotypic predictions, these phenomic insights serve as a powerful validation tool, reinforcing confidence in selection decisions and revealing the underlying biological drivers of performance.

By bridging imagery, AI, and genetics, Hiphen’s approach marks a major step toward data-driven breeding pipelines, where every image collected becomes a predictive layer in accelerating genetic gain and optimizing breeding outcomes across environments.

This use case has been produced from the work of INRAE Le Moulon within their Phenomaize experiment.

Evaluating Sunflower seed moisture through drone imagery and digital traits in a non-destructive and standardized way.

Measuring Sunflower seed moisture from the sky.

In sunflower breeding, seed moisture content is much more than a harvest metric — it’s a critical indicator of yield stability, oil quality, and genetic performance under diverse environmental conditions. Yet, assessing it accurately has long relied on destructive, time-consuming sampling methods that limit the number of plots breeders can evaluate.

Through the HELEX Project, in collaboration with Syngenta, MAS Seeds, and Innolea, Hiphen has developed a groundbreaking digital trait that changes the game: sunflower seed moisture estimation directly from drone RGB imagery.

Through Deep Learning models, this innovative trait allows breeders to measure moisture levels of the seeds non-destructively — at scale, across all plots and even across multi-location trials — using the same standardized protocol.

Repeated measurements throughout the crop cycle generate dynamic desiccation curves, describing each genotype’s dry-down kinetics and helping identify varieties that dry faster, more uniformly, and with greater resilience.

Beyond improving selection accuracy, this approach empowers breeding teams to optimize trial resources and harvest logistics, ensuring that data collection and harvesting are perfectly aligned with the physiological maturity of each variety.

Drone phenotyping is redefining precision for sunflower breeders — Get in touch if you’re interested!

Providing Yield Indicators Through Organs Count and Morphology Traits From Drone Imagery Directly in the Field.

Measuring yield components in vegetable trials with drone imagery straight after harvest.

For vegetable breeders, understanding yield components remains one of the most time-consuming steps of the selection process. Once harvest is complete, collecting reliable data on fruit count, size, and shape traditionally requires heavy logistics, specialized equipment, and hours of manual work — often limiting the number of genotypes that can be evaluated.

With PhenoScale by Hiphen, breeders can now access precise, standardized yield-related traits directly from the field — right after harvest. By simply leaving the vegetables visible in their plots and performing one final drone flight, PhenoScale captures high-resolution RGB imagery that is then processed into digital traits describing organ count, morphology, and trait heterogeneity.

Within just a few days, breeders receive quantitative insights that complement yield information — providing a clear view of the productivity, uniformity, and morphology of each tested variety.

These traits help identify genotypes with superior yield potential and better understand the genetic diversity within breeding populations. By simplifying post-harvest characterization, PhenoScale empowers breeders to analyze more plots, more precisely, and more consistently, accelerating the selection of high-yielding vegetable varieties and refining the genetic material that will shape tomorrow’s crops.