Industrial Perception Research

Shrimp grading using a robotic arm and computer vision

A station by station map for Indian processors: the count per pound grade verification pipeline, the vision system specification with parameters and fallbacks, and a 16 point vendor checklist for a shrimp line.

A shrimp grading line with an overhead camera reading count per minute and grade while a robotic arm removes rejects from the belt

A 40 to 60 MT/day Indian shrimp plant runs several hundred workers across deheading, peeling, deveining, grading and packing, QC by sampling and eye. Labour is ₹10,000 to 15,000 per worker per month. At that price, automation that only replaces headcount rarely pays back, and loses money if it costs yield. Raw material is 70 to 80% of cost; margins are low single digits.

The target is not headcount but measurement: recovering lost yield, capturing unrealised price, catching losses (rejections, claims, failed audits) before shipping. This page maps the line, then details the highest payback pipeline: grade verification. The line detail here comes from our pilot work with Danica Aqua Marine, Visakhapatnam.

Where labour replacement automation fails

Automatic deheading lines quote up to 70% lower labour cost and 2 to 3 times manual throughput. A peeling line cuts 100 manual workers to about 40, fewer on a PUD line. Removing 60 workers at ₹12,000/month saves about ₹86 lakh a year.

But automatic peeling yields 3 to 5% below hand peeling, mostly at the tail tip. With raw material at 70 to 80% of cost and margins of 2 to 4%, that loss can exceed the labour saved. So plants stay manual, and the useful vision work measures the process, not replaces it.

What is already sold, and its limit

CategoryExamplesDoesDoes not
Optical sortersTOMRA, Bühler, Key TechnologyEject peel, shell, legs, eyes, tails, by catch from a bulk flow. ₹1.5 cr and up.Verify count per pound, assess incoming lots, attribute yield, per batch audit
Peeling / deveiningLaitram, Gregor JonssonPeel 10 to 90 count, all product forms, up to 10x hand speedLoad themselves (manual), measure anything
Grading / weighingMarelGrade by physical dimension, weigh, batchMeasure count per pound (dimension is a proxy, not the count)
Regional linesYuanyu, Foshan Soontrue, PowertechDeheading and PD lines at low costMeasure the process

Optical sorters eject product from a flow. None measure the process producing it. Retrofit vision does, at ₹5 to 15 lakh, on a line already running and paid for.

The line at a glance

NEEDS PERCEPTION MECHANISABLE SOLVED BY MACHINE 1 2 3 4 5 6 7 8 9 10 11 12 1 Receiving2 Washing 3 Deheading4 Peeling 5 Grading6 Soaking 7 IQF8 Glazing 9 Weigh/pack10 Metal det. 11 Cold store12 Loading
Green: machines already do it well. Amber: mechanisable, but labour based payback is weak in India. Clay: unsolved. It needs perception. The clay dots mark stations that keep a mechanical operation but carry a vision layer on top. Two stations are pure perception and pure margin: Receiving (1) and Grading (5).
StationStatusWhere vision fitsCost
ReceivingPerceptionPer lot size distribution, count per pound estimate and melanosis incidence from a sampling tray, in a 60 second report at the point of purchase₹4 to 8 L
DeheadingMechanisableCut position monitoring: yield loss per table, shift and operator (management data, not machine control)₹5 to 9 L
PeelingMachine solvedLoading the machine is manual; robotic singulation of wet, clumped shrimp is an open research problem, not a product₹35 to 45 L
GradingPerceptionCount per pound verification. The pipeline below.₹12 to 15 L
MelanosisPerceptionPer pack black spot incidence and severity, added as a class on the grading install₹6 to 10 L
Weigh / packWeighing solvedTimestamped, batch indexed audit record for BRC and BAPOn grading install
Metal detectionSolvedSurface plastic and organic debris that X ray misses (complements, does not replace, metal detection)₹4 to 7 L
PalletisingMechanisableSix axis cobot with vacuum tool. Labour based payback, so weak in India₹35 to 45 L

Washing, soaking, IQF freezing, glazing and cold store use standard equipment, no vision needed. The rest of this page is the grading pipeline, the largest and most direct return.

Grading: the highest value case

Price is set by count per pound: bands like 16/20, 21/25, 26/30, a lower number meaning larger shrimp and higher price per kilogram. Mechanical roller and weight graders sort by physical dimension, a proxy for count. The two diverge with moisture, soaking uptake, product form (head on, headless, tail on) and size spread within a band. A pack one band low gives away margin; one band high draws a buyer claim.

The number, not the labour.

At ₹135 to 140 crore annual revenue, a 0.5% grade realisation gap is about ₹68 lakh, roughly a quarter of the year’s profit at a 2% margin. A 0.2% gap is ₹27 lakh. A 2% gap, which is not unusual, erases the year’s profit.

The pipeline

  1. Capture. Overhead global shutter camera (2 to 5 MP, 30 to 60 fps) over the grading discharge or sampling tray, fixed working distance, controlled LED, a colour and scale reference card in frame for per frame calibration.
  2. Instance segmentation. Isolate every shrimp, including touching ones, into per object masks. Model: Mask R-CNN or a YOLO-seg / RTMDet-Ins class network trained on plant imagery. Clumped clusters are flagged and excluded from the count, not guessed.
  3. Morphometry. Per mask: medial axis length with curvature correction (a spline fitted to the skeleton straightens the curl), projected area, and mid body width.
  4. Weight and count. Convert size to individual weight with a calibrated length weight model W = a·Lⁿ or a small learned regressor. Count per pound is 453.6 divided by mean individual weight in grams. The grade band follows, with an explicit interval from quantile regression or an ensemble.
  5. Per pack verdict. Aggregate over a pack or batch window: measured count distribution against the declared band, percent in band, and the direction of any miss.

Deriving weight: 2D or 3D

Sizing accuracy is won or lost here, and the sensor sets the ceiling. Grade is count per pound (weight), not what a mechanical grader measures. Three options, descending in accuracy and cost:

2D morphometry overhead camera curvature corrected length + area length + area + axis to a weight regression weight ± 5 to 8% blind to height and volume 3D: laser profile or RGB D laser line reconstructed volume volume to a weight regression weight ± 3 to 5% handles soaking, multiple forms
Grade is count per pound, which is weight. A 2D top view measures length and area; a 3D profile measures volume, which tracks weight more closely when moisture and soaking change the density.
MethodHow weight is derivedWeight targetUse when
3D: laser profile or RGB DRecover volume, then a volume to weight regression± 3 to 5%Multiple product forms, or soaking in use
2D morphometryInstance segmentation, then length, area and a curvature corrected axis to a per grade weight regression± 5 to 8%A stable single product form
2D length onlyLength against a lookup table, a dimensional proxyCross check onlyVerification against another method, never a primary claim

Sizing is commonly specified as 95% correct grade band assignment. That figure means nothing without its method: ask how it is measured, and against what reference.

Calibration. Coefficients are fit against physically counted and weighed reference trays, never the grader’s own output, which would propagate the error being measured. Recalibration is scheduled, as size distribution and shell condition drift across the culture cycle.

Data required. 20,000 to 30,000 annotated instances across at least five grade bands and both major product forms, plus weighed reference trays for ground truth.

Targets, benchmarked in the proof of concept. Count per pound mean absolute error within about one count for most packs; per pack verdict in seconds; throughput matched to the grading line in shrimp per hour. Numbers to benchmark against, not claimed results.

We have not deployed grade verification in a plant yet. The pipeline and targets above are the design we run a paid, three week proof of concept against, benchmarked pack by pack against your QC team before anything is trusted or acted on.

From real time verdicts to the end of run report

The system produces two things. In the run it emits a verdict per pack in real time: pointers that raise an alert, drive an air jet, or light a screen indication. These are operational, not the deliverable.

The deliverable is a report, finalised at a logical endpoint rather than streamed: a batch or grade change, line stop, or shift end. At that trigger the system closes the window, aggregates its verdicts, reconciles counts against the window’s weighed reference samples, and generates one report per line and batch, readable by an owner, not only a QC head. It carries:

  • Grade realisation. Declared band, measured count distribution, percent in band, percent graded low (margin lost), percent graded high (claim risk).
  • Value in rupees. Low and high grading converted to money against the day’s price sheet. The first line an owner reads.
  • Count per pound histogram per band, showing spread and outliers.
  • Attribution. Which line or grading station is drifting, so it can be reset the same day.
  • Shift trend. Morning against evening, to catch drift from ice melt, fatigue or throughput changes.
  • Defect incidence. Melanosis and foreign material percentages, if those classes are enabled.
  • Batch and lot identity. Which run this is, its declared grade and product form, start and stop times.
  • Throughput and coverage. Pieces assessed, and the percent abstained on for clumping, so the report states its own sampling.
  • Exceptions. Packs the system could not assess, listed rather than silently dropped.
  • Audit record. Every verdict timestamped and batch indexed, exportable for BRC and BAP.

It is generated automatically at the trigger, sent to owner and QC head as a one page summary with the full record attached, and kept as the batch audit trail. Nothing depends on an operator remembering to compile it.

End of day report, illustrative figures
Batch declared grade26/30
Measured mean count (per lb)28.4 ± 1.2
Packs in band82%
Under graded (shipped low)14%
Over graded (claim risk)4%
Value leakage, this batch₹1.9 L
Drift sourceGrader 2
Illustrative only, to show the report’s shape. Real figures come from a proof of concept on your line.

Melanosis: grading black spot, and where we start

Melanosis, the black spot that forms on head and shell within hours of harvest, is enzymatic browning driven by polyphenol oxidase. It is scored 0 to 10 (Otwell and Marshall): 4 is a measurable defect, 8 approaches reject grade. A leading cause of consignment downgrade, with one review putting darkened product at around 10% of export weight. Sampling QC catches the batch average, not the per pack incidence that triggers a buyer claim.

Computer vision melanosis detection on two real shrimp, 36 percent and 7 percent of surface affected, outlined in red
Our CIE Lab colour threshold melanosis detection, run on real shrimp images. Left: heavy black spot, 36% of surface, reject grade. Right: light, 7%. The production system adds learned severity grading on top of this. Images from our research.

The method. Naive RGB thresholding fails: black spot overlaps in RGB with ice glare, wet shell sheen, shadow and natural cephalothorax pigment. What works: illumination normalised CIE Lab colour, texture descriptors, and learned segmentation that grades severity rather than yes or no. The hard part is class imbalance: affected product is often under 2% of throughput, so per class sensitivity, not overall accuracy, is the metric.

Three method tiers, in descending order of what they deliver and demand:

ApproachOutputLimitation
Learned segmentation (Mask R-CNN or YOLO-seg)Per pixel spot masks, a severity score per piece on the 0 to 10 scaleNeeds a substantial labelled dataset
Colour space thresholding (CIE Lab)Percentage of affected surface area, with illumination normalisation and a texture descriptor to separate spot from shadowWeak on early stage spot and on natural dark pigmentation
Binary flaggingPresence or absence at pack level, an audit flagNo severity, no localisation

The data problem. Affected product is often under 2% of throughput, so overall accuracy is meaningless and both training and validation are statistically awkward. Insist on per class sensitivity and precision against expert labelled ground truth, and deliberately source rejected and downgraded lots: a model trained only on good product cannot detect bad product.

Where we start: the data. The scarce input is labelled shrimp imagery under real conditions, which we have been building: a growing corpus from the local Visakhapatnam cluster, hatcheries and farms, spanning size grades, freshness states and melanosis severities. It lets us stand up a melanosis or grading model quickly and validate it against a plant’s own QC, not from zero. The detectors are at proof of concept stage, benchmarked pack by pack before anything is trusted.

Beyond melanosis: the full defect sheet

Melanosis is one visual defect among many. Buyers grade against a defect sheet; a shrimp is rejected for any deviation from spec: attached shell, discolouration, black gill and black head, dropping or loose head, vein not removed, dehydration and freeze burn, clumping, soft shell, deep cuts, broken pieces, foreign matter, and count or size out of grade. Each defect carries a tolerance limit and a class (critical, major or minor); the defect percentage is defective units over sample size, judged against an agreed AQL, critical defects targeted at zero.

A vision layer screens the visual defects on every pack, not one sampled carton, and logs each to the defect sheet automatically. It does not cover smell, decomposition below the surface, or chemical and microbial residues, which stay with the lab and human grader.

For structural defects (broken pieces, deep cuts, loose heads, shell fragments on peeled product, incomplete deveining), three approaches apply, and a new site usually starts with the last:

ApproachBest forNote
Anatomical modellingThe long tail of defect typesInstance segmentation with a shape descriptor compared against a learned model per product form
Contour analysisGross breakagePerimeter, aspect ratio, convexity deficit, segment count; transparent and cheap
Anomaly detectionNew plants with no labelled defect dataTrained only on good product; flags deviation without classifying it

Anomaly detection suits a new site. It needs no defect examples. It flags that something is wrong rather than what, acceptable in a manual assist deployment and running months earlier. Classes differ by line: HOSO work on head and gill discolouration, loose heads and broken antennae; peeled and deveined lines on vein residue, meat tears from the peeler, and tail on completeness.

DefectSurface vision?Note
Count / size out of gradeYesThe core case: per object sizing and count per pound
Melanosis (black spot)YesSeverity on the 0 to 10 scale, per pack
Black gill / black headYesLocalised discolouration at gill and head
DiscolourationYesColour deviation from the spec
Attached shell, shell in PUDYesShell fragments on peeled product
Dropping / loose headYesHead separation on head on product
Deep cut / broken / damagedYesShape and edge break detection
Dehydration / freeze burnYesSurface whitening and texture
Clumping (IQF)YesPieces frozen together
Vein not removedPartlyOnly where the vein shows on the surface
Soft shellPartlyAn appearance cue, not a definitive test
Foreign matterPartlySurface only; sub surface needs X ray
Bad smell / odourNoNot a visual property
Decomposition (early)NoOnly advanced stages reach the surface
Antibiotic / microbial residueNoNeeds a lab assay, not imaging

One camera platform can carry many of these as separate trained classes on one install, so grading, melanosis and the surface defect set run together, not as separate machines. The discipline: claim only the visual set and the count; leave smell, decomposition and residues to the methods built for them.

How to specify the system

Most vision projects on a food line fail for one of three reasons: motion blur at line speed, specular glare from wet product, or a model that validated in commissioning week and drifted by month three. None are vendor specific; all are visible in advance, and are specification failures, not bad luck. Each section below gives the parameter that matters, an indicative target, and the fallback when site or budget rules out the ideal.

Every fallback costs a capability. Stating it puts the compromise in the scope document, not the acceptance meeting.
The values in the tables below are indicative specification targets typical of this class of installation, not measured results from any deployment. Ethan’s grade verification and melanosis detectors are at proof of concept stage, benchmarked pack by pack against a plant’s own QC before anything is trusted or acted on.

Start with the line rate

Every other number follows from this. A line at 500 pieces per minute presents 8.3 per second. If the inspection zone is 400 mm and the belt runs at 0.5 m/s, each piece sits in the field of view for roughly 800 ms: ample in principle, none at all if the camera cannot freeze it. Three measurements precede any specification:

  • Belt speed (m/s), from an encoder or timed marker. Sets the exposure and frame rate requirement.
  • Pieces per minute, counted over 60 seconds three times. Sets the processing throughput.
  • Belt area coverage (%), photographed and estimated. Determines whether clumping is the gating problem.
500 pieces/min = 8.3 pieces/s belt 0.5 m/s field of view 400 mm dwell about 800 ms Motion blur budget at most 1 px along travel else global shutter + strobe
Line rate sets everything downstream. Belt speed and field of view length fix the dwell time; the exposure must be short enough to hold motion blur under about one pixel, or the shutter and lighting have to freeze the motion instead.

Camera and optics

The sensor and lens follow the line rate measured above.

ParameterTargetFallbackWhat the fallback costs
SensorIndustrial global shutter CMOSGlobal shutter at a lower frame rate, belt slowed locallyThroughput, not accuracy
Resolution2 to 5 MP, sized so the smallest defect spans at least 8 to 10 pixelsWider field of view split across two camerasMoney, not capability
Frame rateAt least 60 fps at full resolution30 to 40 fps on a slowed inspection zoneA line modification
LensC mount, focus and aperture mechanically lockedNone: unlocked optics drift within weeks under vibrationFocus and scale calibration
EnclosureIP67 minimum, IP69K preferredIP65 with a secondary shroudWashdown tolerance and service life

Interface. GigE Vision or USB3 Vision; either is standard, the choice following the cable run and controller.

The test that decides shutter type. Capture frames at production speed and measure motion blur. If a shrimp smears more than about one pixel along travel, a rolling shutter will not work and no software recovers it. Rolling shutter works only with strobed illumination short enough to act as the effective exposure, and even then do not claim detection of melanosis spots under 2 mm.

Illumination

Illumination is the item most often underspecified, and it sets the ceiling on everything downstream. Wet chitin is close to a mirror: untreated, specular highlights destroy colour fidelity and edge detection, so the system reads glare as a defect and a shadow as a spot.

AspectTargetFallbackWhat the fallback costs
GeometryEnclosed tunnel or dome, white diffuse inner surfaceOpen diffuse bar with an ambient light shroudColour based classes degrade
SourceHigh intensity LED bars, strobed in sync with the shutterContinuous diffuse LEDEffective exposure lengthens, blur risk returns
Glare controlCross polarisation: a linear filter on both source and lensDiffusion only, no polarisation1 to 2 stops of contrast on wet shell; melanosis severity unreliable
Ambient isolationSealed from hall lightingShroud onlyModel drifts as hall light changes across the shift

Required at every tier: a fixed white and grey calibration reference in the field of view. Near zero cost, and the only practical way to detect and correct illumination drift as lamps age and doors open.

Decision trigger. If specular highlights cover more than roughly 15% of shrimp surface area in your frames, cross polarisation is not optional.

diffuse dome, white inner surface sealed from hall light white + grey reference in frame strobed LED strobed LED polariser (source) polariser (source) camera polariser (lens), crossed specular glare rejected
An enclosed diffuse dome with cross polarisation (a linear filter on both the source and the lens, axes crossed) suppresses the specular glare off wet shell that would otherwise read as a defect. A fixed white and grey card in the frame is the reference that catches illumination drift as lamps age and doors open.

Clumping and singulation

Clumping is the hardest perception problem on the line, and the one most often passed over in a quote. Solve it mechanically first: a vibratory spreader or belt speed differential that separates product before the inspection zone turns a hard problem into an easy one, as established sorting lines rely on. Where that is impossible, two software positions are defensible, both stated for what they are:

  • Confidence gating. The system abstains on heavily overlapped clusters and passes them through rather than guessing. A cluster reported as unassessed is more useful than one guessed at.
  • Top layer assessment. Inspect the visible layer and report coverage percentage with the result. This is a sampled measurement and must be stated as one.

Decision trigger. Above roughly 30% belt area coverage, no software approach is reliable. Either fix the presentation or reduce the claim.

Processing and latency

Inference runs on site, always.

  • Inference latency: at most 15 ms per frame, which sets the reject timing budget.
  • Hardware: an industrial edge module (for example an NVIDIA Jetson Orin class device) or a rugged x86 with an entry GPU.
  • Runtime: ONNX Runtime or TensorRT.
  • Cloud processing: not acceptable for in line decisions; latency and plant connectivity both rule it out.

For detection and report applications (receiving bay lot assessment, audit logging, grade verification without actuation) CPU only inference on commodity hardware is viable at reduced frame rates. That drops the GPU from the bill of materials and puts a camera only system in a far lower capital bracket. Not every application needs an Orin.

Rejection

The rejection mechanism is chosen against the line rate, not the other way round.

MechanismSustained rateActuation timeRequirement
Pneumatic air jet500+ pieces/minAt most 5 ms valve responseClean, dry, oil free air at 5 to 6 bar through food grade filtration
Mechanical paddle or diverter200 to 400/min20 to 50 msNo air
Manual assist (projected light, laser marker or screen)Any rate, operator limitedn/aNo air

Robotic picking is not a rejection mechanism at line speed: a high speed delta cell sustains roughly 150 to 200 picks per minute against a line presenting 500. Robots belong where the task is placement (tray loading, orientation, pack presentation), not rejection; the two should not be conflated in a specification.

Detectinference ≤ 15 ms Encoder trackby belt counts, not a timer Valve firevalve ≤ 5 ms, 5 to 6 bar reject bin detect ≤ 15 ms transport, camera to reject point, tracked by encoder counts valve ≤ 5 ms encoder tracking fires on the right piece, not at the right moment
Rejection is a timing chain: detect within about 15 ms, track the piece by belt encoder from the camera to the reject point, then fire the air valve within about 5 ms. Encoder tracking is what makes it hit the right piece as belt speed varies.

Line integration

Three integration modes, matched to how far a plant will open its line to a supplier:

  • Full PLC integration. Profinet or Modbus TCP, a deterministic timing loop and encoder based belt tracking, for plants willing to open the PLC to a vendor.
  • Digital I/O trigger. The vision controller signals an existing reject controller. This fits most retrofits.
  • Report only. Display, alarm and a logged record, no actuation. The correct mode for a first installation.

Encoder tracking lets the system fire on the right piece rather than at the right moment; without it, reject accuracy degrades as belt speed varies. Most processors will not grant PLC access to a first time supplier, so plan for report only in project one and integration in project two.

Deploying it

One project that automates a whole plant is how automation projects fail. The order that works, none of the early steps needing a robot:

PhaseScopeCost
0Proof of concept: one station, three weeks, accuracy report vs your QC₹2 to 2.5 L
1Permanent grading verification station₹12 to 15 L
1bAdditional stations on the same platform₹8 L each
2Automated rejection: air blast or pusher on the belt₹6 to 10 L
3Vision guided robotic handling or palletising cell₹35 to 45 L

Nothing actuates on a detection until it has run in report only mode for a full season. A model at 88% sensitivity is a useful report; the same model driving a rejector discards good product 12% of the time.

What the plant provides: station access during production, 230V within a few metres, a clamp fixed camera mount (no drilling, IP65), one QC person for benchmarking, permission to record imagery including bad lots, and weighed reference samples. About two days of access over the three weeks, ideally off peak (Andhra cycles peak February to May and August to November).

Drift, revalidation and cross plant generalisation

A model that passes at commissioning can fail quietly later. Known drift sources on a shrimp line: seasonal morphological variation across the culture cycle, lamp ageing and changing ambient light, belt wear altering the background, pond source and feed variation between suppliers, and changes in soaking or treatment practice.

The specification against drift is concrete: an in frame calibration reference, continuous illumination normalisation, an automated drift alarm, and scheduled revalidation each production season against fresh expert labels. Revalidation should be a named line item with a price, not an unbounded obligation in a support contract, for the supplier’s sustainability as much as the buyer’s clarity.

A model trained at one processor routinely underperforms at another: pond source, feed, size distribution, belt speed, lighting geometry and product form all differ. Assume per site fine tuning on two to three shifts of local data over a shared base model, and check it is priced in. A vendor deploying an unchanged base model is describing a demonstration, not an installation.

The vendor checklist

Sixteen questions for a vendor meeting. A vendor who answers all sixteen without hedging is worth talking to. One who quotes a single headline accuracy number and nothing else is not.

  1. Shutter and frame rate. Global or rolling shutter, and what frame rate at our measured line speed?
  2. Motion blur. What is the measured motion blur, in pixels, on our product?
  3. Glare. How is specular glare from wet shell handled: polarisation, diffusion, or neither?
  4. Calibration reference. Is there a calibration reference in the field of view?
  5. Sizing method. Is sizing 3D or 2D, and how is weight derived?
  6. Sizing ground truth. How is sizing ground truth established, and is it independent of our existing grader?
  7. Melanosis output. Is melanosis reported as severity or presence, and against whose grading scale?
  8. Per class metrics. Per class sensitivity and precision, not overall accuracy?
  9. Defect data. How is the defect training set sourced, given defect rates below 2%?
  10. Clumping. What happens when product is clumped: does the system abstain or guess?
  11. Latency. Inference latency per frame, on what hardware, on site?
  12. Rejection. Reject mechanism, valve actuation time, air specification?
  13. Belt tracking. Is belt tracking encoder based?
  14. Revalidation. What is the revalidation schedule, and its cost?
  15. Fine tuning. Is the model fine tuned on our data, or deployed as is?
  16. Enclosure. Enclosure IP rating, and washdown chemical compatibility?

Frequently asked questions

Can computer vision verify shrimp count per pound?
Yes. An overhead camera on the grading line output segments each shrimp, measures length and area, estimates weight, and computes count per pound per pack, flagging packs outside their declared band (16/20, 21/25, 26/30). A mechanical grader sorts by dimension, a proxy that diverges from true count with moisture, soaking and product form.
What computer vision models are used for shrimp grading?
An instance segmentation network (Mask R-CNN or a YOLO-seg / RTMDet-Ins class model) isolates each shrimp, including touching ones. Per mask morphometry gives length (medial axis with curvature correction), area and width. A calibrated length weight model (W = a·L to the power b) or a small learned regressor converts size to weight; count per pound is 453.6 divided by mean individual weight in grams, with an uncertainty interval.
What does a grade verification system report at end of day?
A per batch report: declared grade versus measured count distribution, percent of packs in band, percent graded low and high, the rupee value of that gap against the day’s price sheet, a count per pound histogram, drift by line or station, and a timestamped audit record for BRC and BAP. Threshold breaches raise a same day alert.
Does optical sorting work for shrimp?
Yes. Installed optical sorters such as TOMRA sort warm water shrimp, ejecting peel, shells, legs, eyes, tails and bycatch from a bulk flow at low false reject rates. They do not verify count per pound, assess an incoming lot before purchase, or attribute yield loss to an operator. They eject product; they do not measure the process.
How is melanosis (black spot) detected automatically?
Through colour and texture analysis in a perceptually uniform colour space (CIE Lab) with illumination normalisation, plus learned segmentation that grades severity, not simple RGB thresholding, which confuses black spot with shadow, ice and glare. The difficulty is class imbalance: affected product is often under 2% of throughput, so per class sensitivity is the metric that matters.
What does a vision QC proof of concept cost in India?
A fixed scope proof of concept at one station, about three weeks, with an accuracy report benchmarked against your QC team, is indicatively ₹2 to 2.5 lakh, normally adjustable in full against a later deployment. A permanent grading verification station is roughly ₹12 to 15 lakh, additional stations about ₹8 lakh each.
Is automating shrimp peeling worth it in India?
Often not on labour grounds alone. Automatic peeling runs about 3 to 5% below manual yield, from the tail tip, and with raw material at 70 to 80% of cost and margins in the low single digits, that loss can outweigh the labour saved. The real bottleneck is loading the machine, singulating wet clumped shrimp, a perception and robotics problem.

Every project starts with a proof of concept, usually two weeks to a month, its timing set by how fast you can supply the data and how clean it is. Once it holds up, we build it into a full system around the needs it surfaced.

PDF
Technical brief: shrimp grading with computer vision
The measurement gap, the full defect sheet, and the melanosis detection, on one page.

Working with us

We run paid, fixed scope proofs of concept at one station, three weeks, with defined deliverables and an accuracy report benchmarked against your QC team. The fee is adjustable in full against a later deployment.

Before that, at no cost: send ten minutes of footage from your grading table or receiving bay. We return an analysis of what is detectable on your line, what is not, and what it would take.

Send footage for a free line review
Ethan AI  ·  Computer vision consulting for industrial and food processing  ·  Visakhapatnam  ·  hello@ethanai.in

The platform does real time perception on commodity hardware, edge or CPU inference, no customer GPU, from standard IP cameras over RTSP and ONVIF. Ethan AI built it for behavioural state detection in clinical settings, and part of the proceeds from this consulting work funds that research. Clinical figures do not transfer to shrimp; the architecture and the practice of validating against expert labels do.