GranoScan: an AI-powered mobile app for in-field identification of biotic threats of wheat
Google Photos will soon help you identify AI-generated images
While animal and human brains recognize objects with ease, computers have difficulty with this task. There are numerous ways to perform image processing, including deep learning and machine learning models. For example, deep learning techniques are typically used to solve more complex problems than machine learning models, such as worker safety in industrial automation and detecting cancer through medical research. As the difference between human and synthetic content gets blurred, people want to know where the boundary lies. People are often coming across AI-generated content for the first time and our users have told us they appreciate transparency around this new technology. So it’s important that we help people know when photorealistic content they’re seeing has been created using AI.
If the current bounding box position is within the + or − of threshold (200 pixels), then we can take the previously saved tracking ID and update the existing y1/ x1 and y2/ x2 locations. Otherwise, generate a new tracking ID and save the y1/ x1, and y2/ x2 positions of the bounding box. Before generating a new cattle ID, we check the new cattle position because the newly detected cattle can also be old cattle which was discarded due to missed count reaching the threshold. When this happens then the new Cattle ID is not generated, and the cattle is ignored. Now, researchers from the Qingdao Institute of Bioenergy and Bioprocess Technology (QIBEBT) of the Chinese Academy of Sciences (CAS) and their collaborators have proposed a new system called “EasySort AUTO”that allows single-cell analysis even of bacteria. Training the AI took around a month, using 500 specialist chips called graphics processing units.
The fivefold cross-validation results, with a mean accuracy of 0.95 and precision of 0.95, along with their respective standard deviations of 0.01, provide strong evidence of the proposed model’s robustness and reliability. The consistent performance across different folds suggests that the model is likely to perform well, effectively balancing correctness and precision in identification. Unfortunately, simply reading and displaying the information in these tags won’t do much to protect people from disinformation. There’s no guarantee that any particular AI software will use them, and even then, metadata tags can be easily removed or edited after the image has been created. Many generative AI programs use these tags to identify themselves when creating pictures. For example, images created with Google’s Gemini chatbot contain the text “Made with Google AI” in the credit tag.
The patent is the first and only livestock biometrics patent of its kind, according to the company. BURLINGTON, Vt. (WCAX) – As artificial intelligence rapidly evolves, it’s important to know how to decipher between real and fake while surfing the web. GPTZero is an AI text detector that offers solutions for teachers, writers, cybersecurity professionals and recruiters. Among other things, the tool analyzes “burstiness” and “perplexity.” Burstiness is a measurement of variation in sentence structure and length, and perplexity is a measurement of how unpredictable the text is. Both variables are key in distinguishing between human-made text and AI-generated text.
3 GranoScan performances towards users’ real use
For instance, adding music to an audio clip might confuse the classifier and lower the likelihood of the classifier identifying the content as originating from that AI tool. An alternative approach to determine whether a piece of media has been generated by AI would be to run it by the classifiers that some companies have made publicly available, such as ElevenLabs. Classifiers developed by companies determine whether a particular piece of content was produced using their tool.
She is a research fellow at the MIT Open Documentary Lab, a member of Women+ Art AI, and holds an MFA in Cinema and Television from Tel Aviv University, where she majored in interactive documentary making. Although this piece identifies some of the limitations of online AI detection tools, they can still be a valuable resource as part of the verification process or an investigative methodology, as long as they are used thoughtfully. Similarly, a model trained on a dataset of public figures and politicians may be able to identify a deepfake of Ukraine President Volodymyr Zelensky, but may falter with less public figures like a journalist who lacks a substantial online footprint.
- Classifiers developed by companies determine whether a particular piece of content was produced using their tool.
- Live Science spoke with Jenna Lawson, a biodiversity scientist at the UK Centre for Ecology and Hydrology, who helps run a network of AMI (automated monitoring of insects) systems.
- In the meantime, it’s important people consider several things when determining if content has been created by AI, like checking whether the account sharing the content is trustworthy or looking for details that might look or sound unnatural.
- These characteristics are subsequently inputted into a Support Vector Machine (SVM), which is tightly connected to the final SoftMax layer of VGG16, in order to achieve accurate identification.
- But we’re optimistic that generative AI could help us take down harmful content faster and more accurately.
Like most AI technology, the tools used for medicine have been found to sometimes produce hallucinations — false or even fabricated information and images based on a misunderstanding of what it is looking at. That problem might be partly addressed as the tools continue to learn; that is, as they take in more data and images, and revise their algorithms to improve the accuracy of their analyses. Shirin anlen is an award-winning creative technologist, researcher, and artist based in New York. Her work explores the societal implications of emerging technology, with a focus on internet platforms and artificial intelligence. At WITNESS, she is part of the Technology, Threats, and Opportunities program, investigating deepfakes, media manipulation, content authenticity, and cryptography practices in the space of human rights violations.
OpenAI working on new AI image detection tools
Where \(\theta\)\(\rightarrow\) parameters of the autoencoder, \(p_k\)\(\rightarrow\) the input image in the dataset, and \(q_k\)\(\rightarrow\) the reconstructed image produced by the autoencoder. Visual comparison of ultrasound images from patients with and without PCOS. I’m not overly critical of adding AI tags for sky replacements, adding a moon, adding a dreamy floral background, etc, because people have been doing this for decades. Frankly, it’s ridiculous to have to try to avoid tools in Photoshop, it has been very disruptive to my workflow.
Unlike traditional approaches, Haider et al.31 proposed a method that incorporates full contextual information surrounding the face from the provided dataset. Leveraging InceptionNet V3 for deep feature extraction, they employed attention mechanisms to refine these features. Subsequently, the features were passed through transformer blocks and multi-layer perceptron networks to predict various emotional parameters simultaneously.
Take the false implication case of 61-year-old Harvey Murphy Jr. as an example of facial recognition gone wrong. Murphy was falsely identified as a thief by the facial recognition-powered security systems at Sunglass Hut. He was arrested and imprisoned for two weeks before authorities realized he was innocent. Authorities later learned that Murphy wasn’t even in Texas during the time of the Houston Sunglass Hut robbery. Murphy alleged the assault left him with “lifelong injuries” in a suit against the Sunglass Hut’s parent company, EssilorLuxottica.
These characteristics are subsequently inputted into a Support Vector Machine (SVM), which is tightly connected to the final SoftMax layer of VGG16, in order to achieve accurate identification. The predicted ID and its related tracking ID are carefully recorded in a CSV file, creating a thorough database for determining the final ID in the future. To address the potential presence of unknown cattle, we thoughtfully store additional RANK2 data, ensuring comprehensive coverage of various identification scenarios.
- The incorporation of RGB (Red, Green, Blue) imaging for individual cow identification signifies a point at which technology harmoniously merges with the welfare and efficiency goals of established farming processes.
- Detection tools should be used with caution and skepticism, and it is always important to research and understand how a tool was developed, but this information may be difficult to obtain.
- Regrettably, not all livestock are monitored on a daily basis due to the significant amount of time and work involved.
- Seal haul-out sites are photographed from overhead passes by manned aircraft or drones, and the resulting images are then annotated in the VIAME software to draw bounding boxes.
- With more sophisticated AI technologies emerging, researchers warn that “deepfake geography” could become a growing problem.
Speaking of which, while AI-generated images are getting scarily good, it’s still worth looking for the telltale signs. As mentioned above, you might still occasionally see an image with warped hands, hair that looks a little too perfect, or text within the image that’s garbled or nonsensical. Our sibling site PCMag’s breakdown recommends looking in the background for blurred or warped objects, or subjects with flawless — and we mean no pores, flawless — skin. The SDXL Detector on Hugging Face takes a few seconds to load, and you might initially get an error on the first try, but it’s completely free. It said 70 percent of the AI-generated images had a high probability of being generative AI. The GIP Digital Watch Observatory team, consisting of over 30 digital policy experts from around the world, excels in the fields of research and analysis on digital policy issues.
The datasets employed in this study were obtained from publicly available sources, and their utilization adheres to ethical standards and guidelines. As these datasets are publicly accessible, specific informed consent was not required. Each metric offers a unique perspective on the performance of the model, capturing different aspects of the classification task. Also, the “@id/digital_source_type” ID could refer to the source type field.
To minimize the throughput of requests for tokens management, refresh and access tokens are stored in a specific private area of the mobile app until the time expires. To evaluate the tracking accuracy of the system for Farm A, a total of 71 videos (355min long) in the Morning and 75 videos (375 min long) in the Evening. These videos specifically included cattle and were recorded on the 22nd and 23rd of July, as well as the 4th and 5th of September, and the 29th and 30th of December 2022. Morning and Evening videos of each day contained the total cattle from a range of 56–65. According to the results, there were some ID-switched cattle due to the False Negative from the YOLOv8 detector.
Where TP is the number of correctly identified cattle and Number of cattle is the total number of cattle in the testing video. To determine the final ID for each tracked cattle, we count the appearances of each predicted ID within the region of interest for that cattle. 20 is 10,328 because the total number of is higher than the other predicted IDs in RANK1, the sample result is shown in Fig.
Meanwhile, companies based in the United States — and other countries with weak privacy laws — are creating ever more powerful and invasive technologies. Clearview claimed to be different, touting a “98.6% accuracy rate” and an enormous collection of photos unlike anything the police had used before. This image of a parade of Volkswagen vans parading down a beach was created by Google’s Imagen 3. But look closely, and you’ll notice the lettering on the third bus where the VW logo should be is just a garbled symbol, and there are amorphous splotches on the fourth bus. Google Search also has an “About this Image” feature that provides contextual information like when the image was first indexed, and where else it appeared online.
We plan to evolve the software into companion animals next, meaning dogs and cats for lost and found applications. Accuracy rates for AI detection tools can be as high as 98 percent and as low as 50 percent, according to one paper published by researchers at the University of Chicago’s Department of Computer Science. Copyleaks also offers a separate tool for identifying AI-generated code, as well as plagiarized and modified code, which can help mitigate potential licensing and copyright infringement risks.
During the fine-tuning, the parameters of the weak models are frozen and the linear layer only is trained. Phenotyping was conducted both on monocots and dicots (Supplementary Table S3), selected through the overall list of weeds recognized by GranoScan. Considering the agronomic relevance of these seven weeds, the phenotyping activity was necessary to enhance the number of images, completing those retrieved from the in-field acquisition.
Pay close attention to hands, especially fingers as most current-gen AI models struggle to get that right. There are high chances of unusual distortion of hands, fingers, face, eyes, and hair. Also, look for the background and you might see complex patterns that may seem out of place.
When the training is done, the trained SVM can be used to predict the cattle ID by extracting features from the feature extractor or input image. OpenAI, Adobe, Microsoft, Apple and Meta are also experimenting with technologies that help people identify AI-edited images. Without due care, for example, the approach might make people with certain features more likely to be wrongly identified.
In all the other tasks, where an object detection model is used, the images were first annotated by manually drawing a set of rectangular areas in which particular diseases or damages are visible (Figure 1). Each rectangle is labeled with the classes reported in Table 1 for a total of annotations before data augmentation. Afterward, all the images were resized to 256×256 pixels for leaf, spike and stem diseases. The process of differentiating between black and non-black cattle during testing yielded significant advantages. This separation not only reduced the occurrence of misidentifications for both groups, but also improved the accuracy of identification specifically for black cattle.
Therefore, detection tools may give false results when analyzing such copies. Likewise, when using a recording of an AI-generated audio clip, the quality of the audio decreases, and the original encoded information is lost. For instance, we recorded President Biden’s AI robocall, ran the recorded copy through an audio detection tool, and it was detected as highly likely to be real.
For spike-stem-root classification, the macro-average precision is 0.8 while the macro-average recall is 0.75. From cell to farm level, scientific advances have always led to a better understanding of how various components of the agricultural system interact (Jung et al., 2021). AI proposes important contributions to knowledge pattern classification as well as model identification that might solve issues in the agricultural domain (Lezoche et al., 2020). Computer vision has been utilized to provide accurate, site-specific information about crops and their environments (Lu and Young, 2020). For the tracking, it is used in both generating training dataset and testing for the identification method. ID-switching occurs when the identification system incorrectly predicted the cattle ID, and the cattle is labeled with incorrect ID.
Clearview AI—Controversial Facial Recognition Firm—Fined $33 Million For ‘Illegal Database’ – Forbes
Clearview AI—Controversial Facial Recognition Firm—Fined $33 Million For ‘Illegal Database’.
Posted: Tue, 03 Sep 2024 07:00:00 GMT [source]
Additionally, a 3\(\times\)3 max-pooled version of the input is included in the feature stack. This combined output delivers rich, multi-perspective feature maps to the subsequent convolutional layer, contributing to the effectiveness of the Inception module in applications such as medical imaging, in our case with PCOS ultrasound images. Spike, stem and root diseases have been grouped and classification results are reported in Figure 11B. For these three tasks, the images acquired by GranoScan thanks to the users’ field activity make it possible to obtain a fairly limited number of recognition responses (44). Nevertheless, the overall accuracy of the AI tool is high, yielding a value of 95%.
At a high level, AI detection involves training a machine learning model on millions of examples of both human- and AI-generated content, which the model analyzes for patterns that help it to distinguish one from the other. The exact process looks a bit different depending on the specific tool that’s used and what sort of content — text, visual media or audio — is being analyzed. Ton-That says it is developing new ways for police to find a person, including “deblur” and “mask removal” tools.
This process is effective, but must be done afresh with every new thing the AI needs to identify, otherwise performance can drop. Midjourney has also come under scrutiny for creating fake images of Donald Trump being arrested. Rather than relying on a handout from the royal family, or a brand, or anyone else that has their own interests, it’s all the more important to have photojournalists we can trust not to manipulate photos.
Drawing from this work, Groh and his colleagues share five takeaways (and several examples) that you can use to flag suspect images. ● create a user community capable of promoting good agricultural practices through the use of the GranoScan app. We would like to thank the persons concerned from Kunneppu Demonstration Farm, Sumiyoshi Farm and Honkawa Farm for giving every convenience of the study on the farm and their valuable advice. The region of interest for Farm C is limited to the leftmost 150 pixels and the rightmost position at 1750 pixels. We discard any cattle that have a bounding box height and width of less than 600 pixels and 250 pixels, as these dimensions do not encompass the entire body of the cattle.
They noted that the model’s accuracy would improve with experience and higher-resolution images. Until recently, these manipulations have been the product of photo and video editing programs that add, remove or shift pixels; or slow, speed up or clip out video frames. Each of these edits leaves a unique digital breadcrumb trail and Stamm’s lab has developed a suite of tools calibrated to find and follow them. Stamm’s lab has been active in efforts to flag digitally manipulated images and videos for more than a decade, but the group has been particularly busy in the last year, as editing technology is being used to spread political misinformation. “You cannot completely rely on these tools, especially for sensitive applications,” Vinu Sankar Sadasivan, a computer science PhD student at the University of Maryland, told Built In. He has co-authored several papers highlighting the inaccuracies of AI detectors.
For instance, we built the blur and compression into the generation through the prompt to test if we could bypass a detector. We used Open AI’s DALL-E 2 to generate realistic images of a number of violent settings, similar to what a bystander may capture on a phone. We gave DALL-E 2 specific instructions to decrease the resolution of the output, and add blur and motion effects. These effects seem to confuse the detection tools, making them believe that the photo was less likely to be AI-generated.
Even better, he says, is to have several independent journalists attending the same event so they can check each other. Thiessen notes that while the two real photos had realistic lighting—one ambiently lit by the sun slightly to the right of the cheetah, the other captured with head-on flash—the AI cheetah had an unnatural, strong blue highlight in one eye. Blue isn’t necessarily a strange highlight color, especially outdoors when the eyes reflect the sky, but without any other blue hue in the picture, he says this was a dead giveaway. Though the Princess of Wales later admitted to editing the photograph, experts say most times you should leave image verification to the pros.
