ConductVision Enterprise
ConductVision boxes and counts every bottle and cup on a filling or packing line and, in a pilot, flags each unit with a defect the quality team defines.
7 detectionsConductVision detections on this photo. Photo: Gobierno de Castilla-La Mancha, CC BY-SA 2.0, cropped.
Quality & Defect Inspection transforms standard industrial cameras into automated inspection stations capable of identifying surface defects, measuring dimensional tolerances, and verifying correct assembly at line speed. The system is designed to catch scratches, dents, cracks, discoloration, missing components, and misaligned parts that human inspectors often miss during high-throughput production. Models can be trained on a small set of annotated defect images for a new product line, and the platform supports multi-camera inspection cells covering all surfaces of a part in a single pass.
Outcomes
When a customer reports a defect, the plant has sample records and no image of that unit.
Without the image, the team cannot tell whether the defect left the line or happened in transit.
A bottle that tips over can jam a star wheel or guide rail and stop the line.
Each jam costs cleanup, lost product and a restart.
A counter at the end of the line shows output and nothing about where containers drop out.
Units knocked over, pulled by hand or crushed upstream vanish from the record.
When a filler or capper stops, the stop log gives a time and a code and rarely the cause.
Repeat stops from one cause look like separate events, so the fix waits.
What the model sees
ConductVision detections on this photo. Photo: Elisandra Leite, CC BY-SA 3.0, cropped.
ConductVision detections on this photo. Photo: Gobierno de Castilla-La Mancha, CC BY-SA 2.0, cropped.
Endpoints
Each endpoint is defined before the pilot starts, with what is counted, its unit and how the results are broken down.
| Endpoint | How it is counted | Unit | Broken down by |
|---|---|---|---|
| PrimaryDefect rate | Units flagged with a defect class and confirmed by the line's quality staff, per 1,000 units inspected. | Unit:defects per 1,000 units | By:Line, defect class, shift |
| SecondaryTipped containers | Bottles or cups found lying on their side on the conveyor, counted once per container. | Unit:containers per hour | By:Station |
| Stretch losses | Containers counted into a stretch of line minus those counted out of it, per hour. | Unit:containers per hour | By:Stretch, shift |
| Stops with a cause | Share of line stops that have a saved clip and a cause tag confirmed by a supervisor. | Unit:% | By:Machine, week |
Each pilot measures ConductVision on your own footage against a sample checked by hand, before anyone relies on its numbers.
Flag precision
Share of defect flags that quality staff confirm on the unit or its image, in a sampled set.
% · by defect class, station
Escaped defects
Share of defective units found by the plant's own inspection of sampled units that ConductVision did not flag.
% · by defect class
Container count agreement
Difference between the ConductVision container count and a reviewer's hand count of sampled minutes of footage.
containers per minute (mean absolute difference) · by station
How it works
Connect
Position high-resolution cameras above conveyor belts, at station exits, or within robotic inspection cells. The platform supports area-scan and line-scan cameras and auto-calibrates for part geometry, lighting, and field of view.
Analyze
Each image is processed by a defect segmentation model that localizes anomalies at the pixel level, classifies defect types, and measures dimensional features against CAD-derived tolerances. A reference-comparison engine flags deviations from golden samples.
Act
Parts exceeding defect thresholds trigger reject signals to diverters or robotic arms. Every inspection result, pass and fail, is logged with defect images, classification, and measurements, feeding directly into SPC dashboards and MES systems.
What it detects
Bottle
Each bottle on the conveyor, boxed as it passes the camera
Cup
Cups, tubs and other open containers on the line
Tipped container
A bottle or cup lying on its side, from the shape of its box
Line count
Containers passing each station, counted once per container
Defect classes
The plant's own defect types, trained on its rejected units in a pilot
Features
Use cases
Automotive suppliers must maintain near-zero defect rates across stamped, machined, and painted components. Automated inspection can examine every part at line speed, catching surface defects and dimensional drift that sampling-based QC misses.
Printed circuit board manufacturers need to verify solder joints, component placement, and trace integrity. High-resolution cameras paired with defect models can flag cold solder joints, missing components, and bridging defects before boards reach functional test.
Pharmaceutical lines require that every package has the correct label, lot number, and seal integrity. Vision-based inspection can verify text accuracy, cap torque indicators, and tamper-evident seal condition at full production speed.
pharmamanufacturing
Fabrication shops producing structural steel, pressure vessels, and piping need consistent weld quality across shifts. Camera-based bead inspection can screen every weld for visual defects, routing suspect joints to NDT technicians.
manufacturingconstructionenergy
Food processors must detect foreign objects, color anomalies, and packaging defects before products ship. Vision inspection can identify discolored items, broken seals, and mislabeled packages on high-speed filling lines.
food processingmanufacturing
Manufacturers of glass panels, bottles, and ceramic tiles need to detect chips, bubbles, inclusions, and surface scratches. Line-scan cameras with backlighting can reveal sub-millimeter defects invisible to the naked eye at production speed.
The challenge
Manual visual inspection is subjective and fatiguing, leading to missed defects that reach customers and generate costly returns.
How ConductVision helps
ConductVision can analyze every item on the line against trained quality models, flagging defects such as cracks, dents, discoloration, and foreign particles. Defective items can be diverted automatically, and defect type data can feed into root cause analysis dashboards.
manufacturinglogisticsretail
The challenge
Complex assemblies with many components are prone to missing parts or incorrect configurations that are difficult to catch downstream.
How ConductVision helps
The platform can verify that each assembly matches its reference configuration by checking component presence, orientation, and placement. Deviations can be flagged before the product moves to the next station, preventing costly rework later in the process.
manufacturingautomotivelogistics
The challenge
Surface finish defects like scratches, pitting, and coating inconsistencies are difficult for human inspectors to detect consistently at production speeds.
How ConductVision helps
ConductVision can use high-resolution image analysis to detect surface anomalies in real time. The system can classify defect types and severities, enabling quality teams to trace issues back to specific process parameters or equipment wear.
manufacturingautomotiveconstruction
The challenge
Manual measurement of parts is slow and introduces human error, creating bottlenecks in quality control processes.
How ConductVision helps
The platform can perform non-contact dimensional checks using calibrated camera views, comparing measured dimensions against tolerances. Out-of-spec parts can be flagged immediately, and measurement trend data can enable statistical process control.
manufacturingautomotiveconstruction
The challenge
Incorrect packaging including missing items, wrong quantities, or damaged containers leads to customer complaints and return logistics costs.
How ConductVision helps
ConductVision can inspect packages at the end of the line, verifying item counts, seal integrity, and package condition. Alerts can stop the line or divert non-conforming packages before they enter the distribution chain.
manufacturinglogisticsretail
The challenge
Mislabeled products create regulatory risk and erode customer trust, especially in food, pharmaceutical, and chemical industries.
How ConductVision helps
The platform can read and validate label text, barcodes, and date codes against expected values for each production run. Mismatches can trigger alerts, and all label reads can be logged for traceability and audit purposes.
manufacturinglogisticshealthcare
Technical overview
Accuracy and speed are measured on your footage and cameras during the pilot, and reported with the conditions they were measured under.
Industries
Watches stations, robots and crews across the factory floor.
Counts vehicles and tracks crews and robots in plants and yards.
Counts traffic and watches crews on bridges, rail lines and roads.
Tracks staff and protective gear on assembly lines and in clean rooms.
Checks gloves, masks and goggles, and counts units on filling lines.
Tracks trucks, crews and equipment at sorting halls and landfills.
Related
Reads container codes, wagon numbers, labels and signs from camera frames.
Finds each analog gauge in view, with needle reading calibrated per gauge.
Checks hard hats, vests, goggles, gloves and masks against each zone's rules.
Pilot
A pilot takes one prioritized outcome, runs on your cameras, reports the endpoints above and ends with a results review and a decision.
