AI, Data, Annotation & Innovation — discover the Infoscribe.ai blog
The synergy between human expertise and advanced technologies in the service of data.
Most vision projects that fail had a sufficient budget, a competent team and a performant model. What...
A computer vision quote almost always prices the visible part of the work. The items that make a budget...
The only systematic study devoted to labelling instructions in biomedical image analysis reaches a result...
A recently published corpus audit delivers a result that should change how quality control is designed....
The choice of a medical image segmentation format looks like a minor technical decision and in fact determines...
A study on a thoracic lesion annotation tool publishes a figure that belongs in every costing discussion....
A recent publication states in one sentence what most projects refuse to admit: this variability renders...
A recent study compared eleven open-source tools computing the same distance metrics on the same masks....
A recent publication on a new medical annotation tool states in one sentence the problem most projects...
The lung presents a difficulty few organs share: the boundary that must be drawn often does not exist....
One anatomical detail captures the nature of the cardiac problem. The papillary muscles and trabeculations...
One challenge result says most of what matters about bone segmentation. On vertebral contouring, the...
The organisers of an international tumour segmentation challenge make an admission rarely found in technical...
The debate between two and three dimensional approaches is often settled on principle when it can be...
One recent result summarises the state of the field better than a long exposition. On segmenting abdominal...
This cluster closes on a forward-looking question, and the exercise calls for a methodological precaution....
Demand for weed detection does not come mainly from technical progress. It comes from a regulatory constraint...
The eleven preceding articles in this cluster flagged errors as their subjects arose. This one gathers...
The choice of sensor precedes that of the model and it determines what the system will be able to distinguish....
The cost of a weed detection project distributes differently from that of an industrial vision project,...
Measuring the quality of weed annotation poses a particular difficulty: the reference itself belongs...
A weed detection corpus is not built by collecting images and then having them annotated. The reverse...
Robotic mechanical weeding inverts the trade-off of spraying. A missed weed will be seen again on the...
Targeted spraying is the most widely deployed use of weed detection and the one with the hardest constraints....
Wide-spaced row crops are the most favourable ground for weed detection. The sowing geometry supplies...
Deciding what constitutes a class is the first operation of a weed detection project, and it is the one...
Weed detection is one of the rare computer vision problems where the object to recognise and the object...
This series on satellite image segmentation closes on a forward-looking question, and the exercise calls...
The twelve preceding articles in this series on satellite image segmentation covered classes, modalities,...
The eleven preceding articles in this series on satellite image segmentation flagged errors as their...
The architectures used in satellite image segmentation have passed through three generations in a decade,...
The tooling of a satellite image segmentation project divides into four distinct families, and confusing...
The question of what a satellite image segmentation costs looks simple and it calls for a structured...
The opening article of this series on satellite image segmentation announced that most annotator disagreement...
Hyperspectral imagery is the third modality in this series on satellite image segmentation, after multispectral...
The five preceding articles in this series on satellite image segmentation all concerned what can be...
Linear objects are the fourth case in this series on satellite image segmentation, and they pose a problem...
Vegetation is the third case in this series on satellite image segmentation, and it combines the difficulties...
Buildings present difficulties exactly opposite to those of water in satellite image segmentation. Where...
Water is by some distance the easiest class to detect in satellite image segmentation and one of the...
The nomenclature is the most consequential decision in a satellite image segmentation project and it...
This cluster closes on a forward-looking question, and the exercise calls for a precaution. The fourteen...
Model assistance is presented as the answer to the cost of video annotation. It delivers a real gain,...
A video corpus is not built by accumulating hours of footage. What determines its value is the number...
A sequence containing people is personal data before it is a corpus. That qualification precedes any...
The preceding articles set out the methods. This one examines what no method fully resolves: the configurations...
A video corpus can be correct frame by frame and unusable as a whole. That possibility defines the particular...
The choice of a video tool weighs more on a project than the choice of an image tool. Working throughput,...
The cost of a video project does not follow from its duration. A minute can cost a few euros or several...
This series on aerial LiDAR closes on a forward-looking question, and the exercise calls for a methodological...
The twelve preceding articles in this series on aerial LiDAR flagged errors as their subjects arose....
Aerial LiDAR measures a geometry and imagery describes an appearance. Those two pieces of information...
Point density is the first parameter of an aerial LiDAR survey and the most often misunderstood. It is...
The cost of an aerial LiDAR project distributes differently from that of an imagery project, and the...
Measuring the quality of a point cloud classification poses a problem imagery does not encounter. There...
Annotating a point cloud does not resemble annotating an image. The annotator does not look at a view,...
Several countries have now launched aerial LiDAR coverage programmes at national scale. That development...
The territorial digital twin is the aerial LiDAR use with the broadest ambition and the most heterogeneous...
Vegetation management under power lines is the aerial LiDAR use where geometric precision matters most....
Forestry is the historical application of aerial LiDAR and the one where its value is clearest. It exploits...
Three models derive from a classified aerial LiDAR survey, and confusing them is one of this field’s...
Classifying a point cloud is the core of work on aerial LiDAR data. It turns a set of geometric measurements...
Aerial LiDAR and photogrammetry both produce georeferenced three-dimensional point clouds. That similarity...
Aerial LiDAR measures distances. That is a difference in kind from imagery, which records light intensity,...
The volume of data described by humans has been falling for several years while total spending keeps...
Behind every corpus are people who looked at data, one item at a time, for weeks. What that work costs...
Not all description errors are equal. Some are corrected in a day, others are discovered months later...
Manufacturing data rather than acquiring it solves an availability problem by introducing another. A...
A labeling budget is spent on data that was chosen. Random sampling, which remains the common practice,...
Describing what a piece of data holds and judging what a model has produced are two distinct kinds of...
A labeling instruction is judged not on its precision but on its reproducibility. Two people reading...
Entrusting data to an annotation provider means circulating it among dozens of people, across as many...
A quality assurance arrangement is judged on one thing alone: what it makes visible and what it lets...
A labeling budget is rarely built from measurements. It is deduced from a rate found elsewhere, multiplied...
Spreading annotation work across thousands of anonymous contributors solves a capacity problem by creating...
The annotation movement is learnt in an hour. What separates an experienced annotator from a beginner...
Building an in-house team or entrusting the work to a provider is often decided by setting an hourly...
Describing an object in an image, an intention in a sentence and a speaker in a recording pass for distinct...
A model learns what it is shown. Data labeling is the work that decides what it will see, and that work...
The rectangle is declared obsolete with every generation of models. It nonetheless goes on occupying...
Adding depth to a rectangle looks like a natural extension. It is in fact a change of nature: the drawing...
Having a model propose the boxes before an operator checks them looks like an obvious gain. The gain...
An annotation instruction is judged on one criterion alone: two operators who read it must draw the same...
Not all annotation errors are equal. Some dissolve into the volume and disappear, others lastingly shape...
An annotation quotation often fits on one line: a price, a unit, a volume. That simplicity hides five...
Annotating a video means annotating images that resemble each other. The difficulty comes not from the...
Four numbers are enough to describe a rectangle. There are nonetheless several incompatible ways of writing...
A scene holding three hundred objects a few pixels across is not handled like a scene holding ten. The...
An object half hidden raises a question its shape does not answer. Should the box enclose what is seen,...
An object lying diagonally defeats the rectangle. The box aligned on the image axes surrounds it without...
Two annotators looking at the same object draw two different rectangles. The gap rarely comes from carelessness,...
The choice between a box, a polygon and a mask presents itself as a trade-off between precision and budget....
The bounding box is the simplest geometry an annotator can draw. It is also the one that produces the...
A segmentation corpus is not sized in number of images. What determines what a system will be able to...
A single overlap figure says a corpus diverges from a reference, without saying where or why. Yet the...
The cost of a segmentation project follows neither from the number of images nor from the number of objects....
No posts found