From hallucinations to {hardware}: Classes from a real-world pc imaginative and prescient challenge gone sideways

Laptop imaginative and prescient initiatives not often go precisely as deliberate, and this one was no exception. The concept was easy: Construct a mannequin that would have a look at a photograph of a laptop computer and establish any bodily harm — issues like cracked screens, lacking keys or damaged hinges. It appeared like an easy use case for picture fashions and enormous language fashions (LLMs), nevertheless it shortly became one thing extra sophisticated.

Alongside the best way, we bumped into points with hallucinations, unreliable outputs and pictures that weren’t even laptops. To unravel these, we ended up making use of an agentic framework in an atypical approach — not for activity automation, however to enhance the mannequin’s efficiency.

On this put up, we’ll stroll by way of what we tried, what didn’t work and the way a mix of approaches ultimately helped us construct one thing dependable.

The place we began: Monolithic prompting

Our preliminary strategy was pretty commonplace for a multimodal mannequin. We used a single, giant immediate to cross a picture into an image-capable LLM and requested it to establish seen harm. This monolithic prompting technique is straightforward to implement and works decently for clear, well-defined duties. However real-world information not often performs alongside.

We bumped into three main points early on:

• Hallucinations: The mannequin would generally invent harm that didn’t exist or mislabel what it was seeing.
• Junk picture detection: It had no dependable option to flag photos that weren’t even laptops, like photos of desks, partitions or folks sometimes slipped by way of and obtained nonsensical harm experiences.
• Inconsistent accuracy: The mixture of those issues made the mannequin too unreliable for operational use.

This was the purpose when it grew to become clear we would want to iterate.

First repair: Mixing picture resolutions

One factor we seen was how a lot picture high quality affected the mannequin’s output. Customers uploaded all types of photos starting from sharp and high-resolution to blurry. This led us to check with analysis highlighting how picture decision impacts deep studying fashions.

We educated and examined the mannequin utilizing a mixture of high-and low-resolution photos. The concept was to make the mannequin extra resilient to the big selection of picture qualities it could encounter in apply. This helped enhance consistency, however the core problems with hallucination and junk picture dealing with continued.

The multimodal detour: Textual content-only LLM goes multimodal

Inspired by latest experiments in combining picture captioning with text-only LLMs — just like the method coated in The Batch, the place captions are generated from photos after which interpreted by a language mannequin, we determined to provide it a attempt.

Right here’s the way it works:

• The LLM begins by producing a number of potential captions for a picture. 
• One other mannequin, referred to as a multimodal embedding mannequin, checks how effectively every caption matches the picture. On this case, we used SigLIP to attain the similarity between the picture and the textual content.
• The system retains the highest few captions primarily based on these scores.
• The LLM makes use of these prime captions to jot down new ones, attempting to get nearer to what the picture truly exhibits.
• It repeats this course of till the captions cease enhancing, or it hits a set restrict.

Whereas intelligent in concept, this strategy launched new issues for our use case:

• Persistent hallucinations: The captions themselves generally included imaginary harm, which the LLM then confidently reported.
• Incomplete protection: Even with a number of captions, some points have been missed completely.
• Elevated complexity, little profit: The added steps made the system extra sophisticated with out reliably outperforming the earlier setup.

It was an attention-grabbing experiment, however finally not an answer.

A inventive use of agentic frameworks

This was the turning level. Whereas agentic frameworks are often used for orchestrating activity flows (suppose brokers coordinating calendar invitations or customer support actions), we questioned if breaking down the picture interpretation activity into smaller, specialised brokers would possibly assist.

We constructed an agentic framework structured like this:

• Orchestrator agent: It checked the picture and recognized which laptop computer elements have been seen (display, keyboard, chassis, ports).
• Part brokers: Devoted brokers inspected every part for particular harm varieties; for instance, one for cracked screens, one other for lacking keys.
• Junk detection agent: A separate agent flagged whether or not the picture was even a laptop computer within the first place.

This modular, task-driven strategy produced way more exact and explainable outcomes. Hallucinations dropped dramatically, junk photos have been reliably flagged and every agent’s activity was easy and centered sufficient to manage high quality effectively.

The blind spots: Commerce-offs of an agentic strategy

As efficient as this was, it was not good. Two primary limitations confirmed up:

• Elevated latency: Working a number of sequential brokers added to the whole inference time.
• Protection gaps: Brokers might solely detect points they have been explicitly programmed to search for. If a picture confirmed one thing surprising that no agent was tasked with figuring out, it could go unnoticed.

We would have liked a option to stability precision with protection.

The hybrid answer: Combining agentic and monolithic approaches

To bridge the gaps, we created a hybrid system:

1. The agentic framework ran first, dealing with exact detection of identified harm varieties and junk photos. We restricted the variety of brokers to probably the most important ones to enhance latency.
2. Then, a monolithic picture LLM immediate scanned the picture for anything the brokers may need missed.
3. Lastly, we fine-tuned the mannequin utilizing a curated set of photos for high-priority use instances, like regularly reported harm eventualities, to additional enhance accuracy and reliability.

This mixture gave us the precision and explainability of the agentic setup, the broad protection of monolithic prompting and the arrogance enhance of focused fine-tuning.

What we discovered

A couple of issues grew to become clear by the point we wrapped up this challenge:

• Agentic frameworks are extra versatile than they get credit score for: Whereas they’re often related to workflow administration, we discovered they may meaningfully enhance mannequin efficiency when utilized in a structured, modular approach.
• Mixing totally different approaches beats counting on only one: The mixture of exact, agent-based detection alongside the broad protection of LLMs, plus a little bit of fine-tuning the place it mattered most, gave us way more dependable outcomes than any single methodology by itself.
• Visible fashions are vulnerable to hallucinations: Even the extra superior setups can soar to conclusions or see issues that aren’t there. It takes a considerate system design to maintain these errors in test.
• Picture high quality selection makes a distinction: Coaching and testing with each clear, high-resolution photos and on a regular basis, lower-quality ones helped the mannequin keep resilient when confronted with unpredictable, real-world images.
• You want a option to catch junk photos: A devoted test for junk or unrelated photos was one of many easiest modifications we made, and it had an outsized affect on total system reliability.

Closing ideas

What began as a easy concept, utilizing an LLM immediate to detect bodily harm in laptop computer photos, shortly became a a lot deeper experiment in combining totally different AI methods to deal with unpredictable, real-world issues. Alongside the best way, we realized that a few of the most helpful instruments have been ones not initially designed for the sort of work.

Agentic frameworks, usually seen as workflow utilities, proved surprisingly efficient when repurposed for duties like structured harm detection and picture filtering. With a little bit of creativity, they helped us construct a system that was not simply extra correct, however simpler to know and handle in apply.

Shruti Tiwari is an AI product supervisor at Dell Applied sciences.

Vadiraj Kulkarni is a knowledge scientist at Dell Applied sciences.