How the DJI Thermal Analysis Tool Actually Works
The DJI Thermal Analysis Tool is a desktop application from DJI that takes raw thermal data from their enterprise drones and lets you do radiometric analysis on it. It supports .SEQ and .ARW files from devices like the Matrice 30/30T series and the H30 series payload. You open a video or image sequence, pick regions, and it gives you temperature readings, histograms, and heat maps. That is the short version. The real work comes after you figure out the settings. I installed this on a Windows 11 machine with a dedicated GPU, which matters more than you would think. The rendering pipeline for longer videos eats RAM and VRAM. Once installed, you launch it, import a file, and set your environmental parameters before you do anything else. Emissivity, reflected temperature, distance, humidity, and atmospheric temperature. These are not optional. DJI defaults to a broad range, which is why beginners keep getting readings that are off by 3 to 5 degrees Celsius. The workflow is straightforward but slow if you are not careful. Import the .SEQ file from your drone. Switch to radiometric mode. Calibrate the scene by setting your parameters. Pick your measurement points or draw regions. The tool overlays temperatures in real time. Export your data as CSV or PDF. A typical inspection of a medium-sized electrical panel takes about twelve minutes once you know what you are doing. Before that, it took me closer to forty-five minutes because I kept second-guessing my emissivity values.
One thing most guides skip: the tool does not auto-detect object material. You have to know whether you are looking at oxidized copper, painted metal, or exposed wire. Emissivity for oxidized copper sits around 0.72. Painted surfaces vary wildly depending on color and composition. If you leave emissivity at the default 0.95, which is fine for organic materials like skin or foliage, your metal readings will be wrong. I learned this the hard way during a solar farm inspection where my temperatures were consistently two degrees too low across an entire inverter array.
What the tool actually measures and what it cannot do
The DJI Thermal Analysis Tool processes radiometric data. That means each pixel in your video or image contains an actual temperature value derived from the infrared sensor. It is not just a color map overlaid on visible light. The H30T and similar payloads use a microbolometer sensor, usually with a resolution between 640 by 480 and 384 by 288 pixels depending on the mode. The tool reads that embedded data and lets you query it. Here is the catch. The tool can only analyze what the drone's sensor captured. If your flight altitude was too high, your ground sample distance gets worse and small hot spots disappear into the pixel average. At 80 meters AGL with an H30T in high-resolution mode, one pixel covers roughly 3 centimeters on the ground. A small overloaded connection might only take up two or three pixels. Your temperature reading becomes an average of those pixels, which underreports the actual peak temperature. I ran into this during a distribution transformer inspection. The thermal anomaly was real, but the tool showed a max temperature of 68 degrees Celsius when an independent contact thermometer later read 84 degrees. The difference came down to pixel resolution and the tool's span algorithm smoothing the reading. I had to fly a lower altitude pass and then use the region of interest function to manually pinpoint the hottest pixel instead of relying on the automatic max detection.
Get the Full Details
Another limitation is that the tool does not handle multi-temporal analysis well. If you want to compare temperature readings from three different inspections of the same asset, you have to manually line up the images and record the values yourself. There is no automated baseline comparison feature. DJI's software ecosystem for that kind of work sits in their cloud-based platform, not in the desktop tool. If your job requires trend analysis over months of inspections, plan for extra manual work.
Common mistakes that waste your time
The first mistake is skipping calibration before every flight session. The FLIR sensors in DJI payloads drift. DJI includes an internal calibration shutter that fires between missions, but ambient temperature changes during the day still affect readings. If you start a morning inspection at 12 degrees Celsius and finish at 3:00 PM when the air is 28 degrees, your midday readings will be offset unless you recalibrate. The tool has a calibrate button. Press it. It takes three seconds. The second mistake is ignoring reflection sources. Thermal cameras see reflected radiation, not just emitted radiation. A hot pipe nearby can reflect off a shiny surface and throw off your reading by several degrees. I wasted an afternoon diagnosing what I thought was a failing bearing on a motor because the sun was reflecting off a nearby steel beam into my thermal camera. Moving to a shaded angle and adjusting my position fixed the reading immediately. The tool cannot tell the difference between emitted and reflected energy. You have to. A third mistake is exporting without checking your parameter consistency. The tool lets you save exported reports, but if you changed your distance or emissivity setting between measurements, your report is misleading. I built a habit of writing down my environmental parameters in a notebook before I start each flight. It sounds excessive. It has saved me from submitting incorrect data twice.
When this tool is not the right answer
If you are doing high-volume commercial inspections, the DJI Thermal Analysis Tool is fine for spot checks, but it will slow you down. The manual workflow does not scale. For batch processing dozens of flights, companies typically move to software like FLIR Tools, Tau2, or custom Python pipelines using the raw sensor data. These tools support automated detection algorithms and can process hundreds of images in the time it takes to manually analyze ten in the DJI app. The DJI Thermal Analysis Tool is also not useful if you are working with non-radiometric footage. The free Flycart and other consumer-grade DJI drones do not embed temperature data in their videos. The tool will open the file but you will only get visual analysis, not actual temperature readings. Make sure your hardware supports radiometry before you rely on this software for quantitative work. For professional use, I recommend pairing the tool with a contact temperature reference. Bring an infrared thermometer or a thermocouple to your site. Take a contact reading on a known surface and compare it to your thermal data. This gives you a ground truth check that catches sensor drift and parameter errors before you leave the site. It takes two minutes and has prevented me from filing incorrect reports more times than I can count.
