What Is Yard Intelligence? Computer Vision for Grain Facilities

If you manage a grain elevator or processing facility in Western Canada, you already know what harvest congestion looks like. A line of trucks snaking down the approach road. Drivers idling with their engines running. Your pit operator relaying queue estimates over the radio while trying to keep the dump running. Meanwhile, your office phone is ringing with producers asking if it is worth the drive.

You have likely tried whiteboards, two-way radios, and manual tally sheets to keep tabs on the yard. They work — until they don't. The moment things get busy is exactly the moment you lose visibility.

Yard intelligence is a different approach. It uses cameras you may already own and computer vision software to give you a continuous, automated read on what is happening in your yard — truck counts, wait times, lane utilization, and congestion trends — without anyone having to stand outside with a clipboard.

This article explains what yard intelligence actually is, how the underlying computer vision technology works at a grain facility, what data it produces, and why it matters for day-to-day operations and long-term capacity planning.

What Yard Intelligence Means in Practice

Yard intelligence is real-time visibility into the physical state of your facility yard, generated automatically by software rather than manual observation. At its core, the concept is straightforward: cameras watch the yard, algorithms interpret what they see, and the results show up on a dashboard you can check from your office, your phone, or your truck cab.

The term covers a specific set of capabilities:

• Vehicle detection and counting — knowing how many trucks are on-site at any given moment.
• Queue measurement — understanding how deep the line is at each receiving lane or pit.
• Wait time estimation — calculating how long a truck entering the yard right now can expect to wait.
• Throughput tracking — measuring how many loads per hour are actually moving through the facility.
• Congestion prediction — using current trends to flag problems before they fully develop.

None of this requires a person watching a screen. The system runs continuously, 24 hours a day, seven days a week, through harvest, through overnight receiving, and through the quiet shoulder seasons when you are planning next year's capacity.

How Computer Vision Works at a Grain Facility

Computer vision is a branch of artificial intelligence that trains software to interpret images and video the way a human would — identifying objects, tracking movement, and understanding spatial relationships. In a grain facility context, this means teaching a model to recognize trucks, trailers, and queuing patterns from overhead or angled camera feeds.

Camera Placement

Effective grain facility monitoring starts with camera positioning. The most useful vantage points are typically:

• Approach road / entrance — captures every vehicle entering and exiting the site, establishing a baseline count.
• Receiving lanes and pit areas — monitors queue depth at each active dump point.
• Scale house area — tracks the weigh-in and weigh-out flow, identifying bottlenecks in the ticketing process.
• Staging / holding area — shows how many trucks are waiting for assignment versus actively in line.
• Exit road — confirms departures and closes the loop on per-vehicle dwell time calculations.

Most modern grain elevators already have IP cameras installed for security purposes. Yard intelligence systems are designed to work with existing camera infrastructure. You do not need to rip out your current setup and start over. If your cameras can stream video over your network, they are likely candidates.

What the Software Actually Does

The computer vision model processes each video frame (or a sampled subset of frames — there is no need to analyze every single one) and performs several tasks in sequence:

1. Object detection — the model identifies and draws a bounding box around each vehicle in the frame. It distinguishes between trucks, trailers, passenger vehicles, and equipment.
2. Tracking — once a vehicle is detected, the system assigns it a temporary ID and follows it across successive frames. This is how it knows truck #47 moved forward three positions in the last ten minutes rather than counting three new trucks.
3. Zone mapping — the facility yard is divided into virtual zones (entrance, lane 1, lane 2, scale, exit) during initial setup. The system counts how many vehicles occupy each zone and for how long.
4. Metric calculation — from the raw detection and tracking data, the software calculates derived metrics: average wait time, vehicles per hour, lane utilization percentage, and trend direction.

This entire pipeline runs in near real-time. The delay between a truck pulling in and the dashboard updating is typically measured in seconds, not minutes.

Accuracy and Edge Cases

A fair question from any facility manager: how accurate is this, really? Modern object detection models trained on grain facility imagery achieve detection accuracy above 95% under normal conditions. Performance can dip in heavy fog, blowing snow, or when cameras are obscured by dust buildup — but these are addressable with camera maintenance, IR-capable cameras for low-light conditions, and model tuning for regional weather patterns common across the Prairies.

The system is also trained to handle partial occlusion — when one truck blocks part of another from the camera's view. By combining data from multiple camera angles and using tracking continuity (if a truck existed three seconds ago and didn't leave, it still exists), the model maintains an accurate count even in a crowded yard.

What Data Yard Intelligence Provides

Raw truck detection is the foundation, but the real value is in the metrics that get calculated on top of it. Here is what a facility operator can expect to see:

Truck Counts (Real-Time and Historical)

The most basic metric: how many trucks are on-site right now. This number updates continuously and can be broken down by zone — how many are in line at pit 1, how many are on the scale, how many are in the staging area. Historical counts let you see patterns: Tuesdays are consistently busier than Mondays, the 10 a.m. to 2 p.m. window drives 60% of your daily volume, or a particular week in September is always your peak.

Wait Times

By tracking each vehicle from entry to dump (or from queue entry to service), the system calculates actual wait times — not estimates, not averages from last year, but the measured experience of trucks flowing through your facility today. This metric is valuable both internally (for staffing and lane management) and externally (for communicating honest expectations to incoming producers).

Lane Utilization

If you have three receiving lanes, are all three running? Is lane 2 consistently underused because producers default to the lane closest to the entrance? Lane utilization data shows you where capacity is sitting idle and where it is maxed out, enabling better signage, better traffic flow design, and better staffing allocation.

Throughput Rate

Trucks per hour is the fundamental measure of facility performance. Yard intelligence gives you this number in real-time and over any historical window. You can compare throughput across shifts, across operators, across different commodity types (canola may dump faster than peas, for instance), and across different weather conditions. Over time, this data becomes the basis for evidence-based capacity planning rather than gut feel.

Congestion Predictions

This is where yard intelligence moves from reporting to anticipation. By analyzing the current inflow rate, current service rate, and queue depth, the system can project forward: at the current pace, the yard will be at capacity in 45 minutes. That warning gives you time to act — open another lane, call in a second pit operator, or push a notification to scheduled deliveries suggesting they delay their arrival by an hour.

How Yard Intelligence Integrates with Scheduling

Yard intelligence is most powerful when it is connected to the scheduling layer — the system that assigns delivery windows, manages appointments, and coordinates logistics. Standalone yard monitoring tells you what is happening. Integrated yard intelligence lets you do something about it. (For more on how scheduling works in practice, see how smart scheduling replaces phone calls and our guide on reducing truck wait times.)

Here is how the integration works in practice:

• Dynamic rescheduling — when the yard intelligence system detects congestion building (say, queue depth has doubled in the last 30 minutes), it can trigger the scheduling system to push upcoming appointments back by 30 to 60 minutes, spreading the load before it becomes a problem.
• Slot capacity adjustment — if throughput data consistently shows that a particular lane handles 8 trucks per hour rather than the 10 that were assumed, the scheduling system can reduce the number of appointments offered for that lane, preventing overbooking.
• Producer notifications — when a scheduled delivery window is approaching and the yard is currently backed up, an automated message can advise the driver to delay departure. This is far more effective than having the driver show up and idle.
• Post-season calibration — at the end of a harvest season, the accumulated yard data provides a factual basis for adjusting next year's scheduling templates. Instead of guessing at slot duration and concurrency limits, you set them based on observed throughput.

GrainFlow is built to support this kind of integration. The platform's scheduling engine can consume yard intelligence data to dynamically adjust capacity, notify drivers, and refine delivery windows based on what is actually happening on the ground — not just what the plan assumed would happen.

Practical Benefits for Facility Operators

Technology articles often lean toward the theoretical. Here is what yard intelligence actually changes in the daily work of running a grain facility:

No More Manual Counting

Every facility has some version of the truck count process — someone walks outside, counts the line, radios it in. During peak harvest, this might happen every 15 minutes. Yard intelligence replaces that entirely. The count is always current, always available, and doesn't depend on someone remembering to check.

24/7 Monitoring Without 24/7 Staff

Many facilities run extended hours during harvest but don't have management on-site around the clock. Yard intelligence provides continuous oversight. If a queue builds at 2 a.m. during overnight receiving, the system flags it. If a lane goes idle because of an equipment issue, you'll know from the throughput drop before the night operator thinks to call.

Works With Existing IP Cameras

This is a practical concern that matters: you do not need to install a new camera network. If your facility has IP-based security cameras — and most modern facilities do — the yard intelligence software connects to those existing feeds. Initial setup involves mapping camera views to yard zones and calibrating the detection model for your specific layout. It is not a forklift upgrade. It is a software layer on infrastructure you have already paid for.

Historical Trends for Capacity Planning

The day-to-day value of yard intelligence is operational awareness. The long-term value is the dataset it builds. After one full harvest season, you have a detailed record of every peak, every bottleneck, and every throughput pattern at your facility. This data directly informs capital investment decisions: do you need a third receiving lane, or do you just need better scheduling? The data will tell you.

Safety and Compliance Awareness

Computer vision can be extended beyond truck counting. The same camera feeds can support detection of grain spills near pit areas, excessive dust generation (a real safety and environmental concern), unauthorized vehicles in restricted zones, and pedestrian presence in active truck lanes. These are not primary use cases for yard intelligence, but they are practical extensions that facility safety managers will appreciate.

Reduced Producer Frustration

Long wait times are the number one complaint from producers delivering grain. Most of the frustration comes not from the wait itself but from the uncertainty — driving 45 minutes to the elevator only to find a two-hour line. When yard data feeds into producer-facing tools (like a driver app showing current wait estimates), producers can make informed decisions about when to haul. This reduces peak congestion, shortens actual wait times, and improves the relationship between the facility and its producer base.

What It Takes to Get Started

Adopting yard intelligence is not an all-or-nothing proposition. A practical rollout typically looks like this:

1. Audit your cameras — inventory the IP cameras you already have. Identify which ones have useful views of entry points, receiving lanes, and staging areas. Note the camera models, resolution, and whether they support RTSP streaming.
2. Start with one metric — truck count at the entrance is the simplest starting point. It requires a single camera and gives you immediate value: a running total of daily arrivals and a real-time on-site count.
3. Add queue depth at your busiest lane — once the entrance camera is running, add a receiving lane camera. Now you have both inflow data and service data, which together produce wait time estimates.
4. Connect to scheduling — when you have confidence in the data (typically after a few weeks of operation), connect the yard intelligence feed to your scheduling system so that delivery windows reflect real capacity.
5. Expand coverage — add cameras and zones as the value becomes clear. Most facilities reach full yard coverage within a single season.

The total cost depends on your existing camera infrastructure, but for facilities that already have IP cameras, the incremental cost is primarily software. There is no heavy hardware installation, no concrete work, and no disruption to active receiving operations.

The Bottom Line

Yard intelligence is not a futuristic concept. It is a practical application of proven computer vision technology to a problem that every grain facility in Western Canada deals with: knowing what is happening in the yard without relying on manual observation, radio calls, and best guesses.

The truck counting technology works. The data it produces — counts, wait times, throughput, utilization, congestion forecasts — is immediately useful for operational decisions and invaluable for long-term grain facility monitoring and capacity planning. And it runs on cameras most facilities already have.

The facilities that adopt this capability will have a measurable advantage: shorter wait times, better producer relationships, smarter capital spending, and fewer harvest-season surprises. The ones that do not will keep counting trucks by hand.