Object Search in Images
Other languages
Object Search in Images
Use this template to search a document's images for the objects you describe and tag every sighting with a label you can search for across the Discover case.
Insert Template puts this prompt into the prompt box. Claira writes each answer to the destination field you choose: a Memo field, since each sighting gets its own tagged line.
What this prompt is for
This template searches every image in a document for the objects you describe and tags each sighting OBJECT FOUND: or OBJECT POSSIBLE:, or answers NO OBJECT FOUND, so one field search finds them all.
Photographs, site inspections and scanned exhibits can hold visual evidence that no text search will find: a vehicle, a piece of equipment, a product, a logo. The tags never change, so searching the field for OBJECT FOUND brings up every document where the object is clearly visible, and OBJECT POSSIBLE brings up the ones a reviewer should look at. Every answer also says how many images were reviewed and what the document is, including the documents with no sighting.
Use it when the object can be described well enough for a stranger to pick it out. Do not use it to find an object that is only named in the text: the extracted text already answers that with an ordinary search. For several unrelated objects, run one scan per object into separate fields, so each stays searchable on its own.
Step-by-step in Claira
- In Prompt Lab, click Quickstart above the prompt box. Bulk Scan has the same button.
- Under Prompt Library, open Images, audio and video. On the Object Search in Images card, click Expand to read the prompt, then click Insert Template.
- If the prompt box already has text, Replace existing prompt? asks what to do with it. Replace keeps the prompt tied to this template; Append adds the template below your text.
- In the prompt box, replace the bracketed placeholder with the object you are looking for and the details that set it apart: color, markings, make or model, damage.
- Set Scan as, next to Scan document, to Image so Claira sees the images themselves. See Media scans for the accepted file formats and which models can run an Image scan.
- Open a document in Nuix Discover and click Scan document. The answer appears under AI Response. To save it, choose a Memo field in Apply to field... and click Apply. Do this on 10 to 25 documents first. Single scans are free.
- When the sample reads well, select the documents in Nuix Discover and click Start bulk scan. In Ready to start bulk scan, check the Destination field, then click Start bulk scan. Then search the field for
OBJECT FOUNDto bring up every clear sighting.
The prompt
Search every image in this document for the following object or objects:
[insert each object to find, with the details that set it apart, such as color, markings, make, model or damage]
Images include photographs, scanned pages, drawings, screenshots and pictures embedded in the document. An object is found when you can see it clearly and it matches the description, including its distinguishing details. It is possible when it may match but the view is partial, blurred or distant, or a distinguishing detail cannot be confirmed. Tag only the objects described above, and only what you can see: a mention in the text is not a sighting.
If the document has no images, or none can be seen clearly enough to judge, answer NO OBJECT FOUND and say why in BASIS.
Answer in exactly this form:
one line for each sighting, starting OBJECT FOUND: for a clear match or OBJECT POSSIBLE: for an uncertain one, then the object's name, the page, photo or image it is in, and one or two sentences on what it looks like and what surrounds it, or for a possible match what makes it uncertain; or the single line NO OBJECT FOUND when none of the objects appears
BASIS: one sentence saying how many images you reviewed and, when nothing was found or a match is uncertain, why
DESCRIPTION: one or two sentences saying what the document is, who made it and when, as far as the document showsRecommended customizations
- Put the distinguishing details in the placeholder. "A yellow forklift with a blue cage" returns fewer false sightings than "a forklift".
- To be told about near misses, add a sentence such as "Tag any forklift of another color as OBJECT POSSIBLE: and give its color."
- Keep
OBJECT FOUND:,OBJECT POSSIBLE:andNO OBJECT FOUNDexactly as written. Field searches, and comparisons with earlier scans in the same Discover case, only work while every answer uses the same tags.
Worked example
The placeholder reads: "a forklift".
Input excerpt
A site inspection report prepared by Okafor Safety Consulting on 9 June 2025 for the Eastfield distribution center, with 12 photographs of its loading dock and storage aisles. Photo 7 shows a warehouse aisle with a yellow counterbalance forklift, forks lowered, parked beside pallet racking stocked with wrapped pallets. Photo 11 shows the loading dock, with the rear of a yellow vehicle cut off at the edge of the frame.Expected output shape
OBJECT FOUND: forklift. Photo 7: a yellow counterbalance forklift parked in a warehouse aisle with its forks lowered, beside pallet racking stocked with wrapped pallets.
OBJECT POSSIBLE: forklift. Photo 11: the rear of a yellow vehicle at the edge of the loading dock photo, too cropped to confirm it is a forklift.
BASIS: 12 photographs reviewed; photo 7 shows a forklift clearly and photo 11 may show one in part.
DESCRIPTION: Site inspection report for the Eastfield distribution center, prepared by Okafor Safety Consulting on 9 June 2025, with 12 photographs of the loading dock and storage aisles.Troubleshooting
- Nothing found in documents that plainly show the object. Check that the scan ran as Image, and that your description matches what the images actually show rather than what the object is called.
- False sightings. Add distinguishing details to the placeholder, or a sentence saying that partial or distant views belong under
OBJECT POSSIBLE:. - A field search misses sightings. Check the tag spelling in a few answers: the search only finds tags written exactly as in the prompt.
NO OBJECT FOUNDwith a BASIS saying there were no images. The file sent to Claira had no images, for example a document whose native file is text only. Check the document in Nuix Discover.
Related
Was this page helpful?
Continue reading
Need more help?
Contact our support team at support@claira.to — we are here to help.