Color Labeling Research Articles

Verbal descriptions improve visual working memory but have limited impact on visual long-term memory.

How do verbal descriptions affect visual memory over the short and long term? Here we show for the first time that verbal labeling can boost visual memories, but the source of this benefit depends on whether representations are maintained over the short term in visual working memory or over the long term in visual long-term memory. Across three experiments, we contrasted color memory of randomly colored objects when participants labeled (a) the color, (b) the object, or (c) the color-object binding, to memory under an articulatory suppression condition inhibiting labeling. Memory was tested at two time points: after three objects (visual working memory) and at the end of the experiment (visual long-term memory). In Experiment 1, color labeling improved, whereas object labeling impaired, visual working memory in comparison to suppression. Visual long-term memory remained unchanged across conditions. Experiment 2 tested whether this was attributable to poor overall long-term learning by repeating the colored objects over three successive working memory trials. This increased performance over the short and long term, yet labeling did not change learning rate over repetitions or delayed memory performance, showing no long-term memory benefit. In Experiment 3, a labeling benefit was observed when the color-object binding was labeled both over the short and long term. Mixture modeling indicated that color-labeling benefits in visual working memory resulted from an increase of detailed visual memory, whereas long-term memory benefits accrued from categorical representations. Our findings point to dissociations on the role of language in visual working memory and visual long-term memory. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Color Communication behind Shades of Beige: An Analysis of Color Labels and Terms of Liquid Foundation

With the increasing awareness of skin diversity, cosmetic companies have created foundations with more than 40 shade options. Color researchers endeavored to evaluate the color quality of foundation from product R&D. At the same time, cosmetics companies have conducted their own trials to systematically communicate these tiny shade differences to consumers. In this study, we addressed the color communication for liquid foundation. Through 1009 color labels collected from 26 product lines, two features of color, lightness and undertone were used commonly to classify the shade colors. Lightness was labeled using numbers. The smaller the number is, the lighter the color is, which is similar to the ways numbers are used to specify lightness for surface color. Undertone, the color nuance independent of lightness, was labeled symbolically through sets of initials, such as P(pink), B(beige), and Y(yellow) or simply C(cool), N(neutral), and W(warm).Furthermore, clear tendencies were observed regarding color name in terms of lightness. The similarities and differences when compared with other perception-based color systems and skin color were also discussed. We expect cosmetics companies and manufacturers to utilize the findings from this study for product research and launching.

Loss of nr4a1 abrogates Fitness of asxl1-mutant Hematopoietic Clones

Clonal fitness of mutant hematopoietic stem and progenitor cells (HSPCs) underlies clonal hematopoiesis (CH), a state of clonal expansion associated with increased risk of blood malignancies and cardiovascular disease. Mechanisms by which acquired mutations lead to clonal fitness are not known. We used a zebrafish model to study the effect of acquired asxl1 mutations on HSPC clonality with TWISTR (Tissue editing With Inducible Stem cell Tagging via Recombination) that combined mosaic CRISPR-Cas9 mutagenesis with color labeling of HSPC clones. TWISTR asxl1 mutants showed clonal dominant states with the expansion of single-colored clones contributing to over 30% of myelopoiesis. These zebrafish had normal hematopoietic output and no major lineage skewing. Single-cell RNA sequencing analysis of TWISTR mutant marrow cells harboring asxl1 mutations showed >10-fold upregulation of inflammatory cytokines in mutant mature myeloid cells and >30-fold upregulation of anti-inflammatory modulators in immature progenitors. Increased inflammation has been widely documented in persons with CH with acquired mutations in DNMT3A, TET2, ASXL1 and other genes. Moreover, chronic inflammation due to infection was shown to promote relative clonal fitness in Dnmt3a mutant mice. Based on our results, we proposed the hypothesis that upregulation of the anti-inflammatory genes, including nr4a1, served as a mechanism of resistance to chronic inflammation created by the mutant HSPC progeny, resulting in a self-perpetuating cycle of clonal fitness in that environment. To test this, we used TWISTR to generate mosaic mutants of asxl1 and nr4a1 by injecting zebrafish with guide RNAs targeting exon 12 of asxl1 and exon 3 of nr4a1 together. Our model would predict that abrogation of nr4a1 expression in asxl1-mutant clones would weaken their fitness relative to clones that maintained nr4a1 expression. We sorted over 300 clones of various sizes from this cohort of zebrafish and sequenced the two targeted genes. We found that clones with frameshift mutations in asxl1 with either no nr4a1 mutations or heterozygous nr4a1 mutations had a clone size of 20%±14% or 19.7%±15% in myeloid cells, respectively. Asxl1-mutant clones with biallelic frameshift mutations in nr4a1 were significantly smaller with a clone size of 13.8±11.5% (p < 0.015). Clones with intact asxl1 did not differ in their clone size independent of nr4a1 genetic status (11.9% wildtype nr4a1 vs 11.5% homozygous nr4a1 mutant). Chemical inhibition of nr4a1 over 3 months resulted in reduced change of edited clones in asxl1-mutant zebrafish compared to vehicle-treated zebrafish, with median change in allelic fraction of 3%±4.8% vs 5.3%±7.5%, respectively. This suggested that upregulation of nr4a1 in asxl1-mutant clones maintains their fitness in inflammatory conditions, potentially by limiting HSPC exhaustion. We successfully used TWISTR to study asxl1 induced CH in zebrafish and identified nr4a1 upregulation as a critical pathway engaged for establishing clonal fitness. DisclosuresZon: Fate Therapeutics: Current equity holder in publicly-traded company, Other: Founder; CAMP4 Therapeutics: Current holder of individual stocks in a privately-held company, Other: Founder; Amagma Therapeutics: Current holder of individual stocks in a privately-held company, Other: Founder; Scholar Rock: Current equity holder in publicly-traded company, Other: Founder; Branch Biosciences: Current holder of individual stocks in a privately-held company, Other: Founder; Celularity: Consultancy; Cellarity: Consultancy.

Stability investigation of the Pix2Pix conditional generative adversarial network with respect to input semantic image labeling data distortion

The peculiarities of image generation by a pretrained conditional generative adversarial network on the basis of semantic scene labeling are investigated. Semantic labeling can be inaccurate and can contain defects that result, for example, from transforming the graphic file formats in which it was stored or transmitted. Cases are discussed of image generation based on such incorrect data—with modification of the hue, saturation, and brightness of the colors in the color labels of various classes of objects. It is determined that changing the hue of the label has an especially strong negative effect on image generation and could result in altering the class of the labeled object. Thus, the distribution uniformity of the label color parameters through the color spaces should be taken into account. Additional requirements should be introduced on the accuracy with which the color labels are represented.

Automated color detection in orchids using color labels and deep learning.

The color of particular parts of a flower is often employed as one of the features to differentiate between flower types. Thus, color is also used in flower-image classification. Color labels, such as ‘green’, ‘red’, and ‘yellow’, are used by taxonomists and lay people alike to describe the color of plants. Flower image datasets usually only consist of images and do not contain flower descriptions. In this research, we have built a flower-image dataset, especially regarding orchid species, which consists of human-friendly textual descriptions of features of specific flowers, on the one hand, and digital photographs indicating how a flower looks like, on the other hand. Using this dataset, a new automated color detection model was developed. It is the first research of its kind using color labels and deep learning for color detection in flower recognition. As deep learning often excels in pattern recognition in digital images, we applied transfer learning with various amounts of unfreezing of layers with five different neural network architectures (VGG16, Inception, Resnet50, Xception, Nasnet) to determine which architecture and which scheme of transfer learning performs best. In addition, various color scheme scenarios were tested, including the use of primary and secondary color together, and, in addition, the effectiveness of dealing with multi-class classification using multi-class, combined binary, and, finally, ensemble classifiers were studied. The best overall performance was achieved by the ensemble classifier. The results show that the proposed method can detect the color of flower and labellum very well without having to perform image segmentation. The result of this study can act as a foundation for the development of an image-based plant recognition system that is able to offer an explanation of a provided classification.

Traffic light signals and healthy food choice: Investigating gender differences

AbstractNutrition labels conveying nutrition information using traffic light signals are becoming common, but traffic light signals' effectiveness to help consumers make healthy food decisions is still debated. In the current research, we propose gender differences in the usage of traffic light signals. In Study 1, we presented male and female participants with a nutrition label that we manipulated as either unhealthy or healthy by changing the nutrient amounts, and the nutrition label either had a green or red color frame surrounding it. Although both men and women responded equally to information conveyed in text form, men were more likely to rely on the color to help them assess the target food product's healthiness. We replicated this result in Study 2: Holistic thinking mediated the men's reliance on color labeling schemes when deciding to buy a targeted food product. These results suggest there are differences in the consumer groups (segmented by gender) for whom traffic light signals on food packaging would be beneficial in playing a role in healthy food decision‐making. For policymakers, our findings indicate that target consumers need to be considered in nutrition labeling generally.

Dual coding of knowledge in the human brain.

How does the human brain code knowledge about the world? While disciplines such as artificial intelligence represent world knowledge based on human language, neurocognitive models of knowledge have been dominated by sensory embodiment, in which knowledge is derived from sensory/motor experience and supported by high-level sensory/motor and association cortices. The neural correlates of an alternative disembodied symbolic system had previously been difficult to establish. A recent line of studies exploring knowledge about visual properties, such as color, in visually deprived individuals converge to provide positive, compelling evidence for non-sensory, language-derived, knowledge representation in dorsal anterior temporal lobe and extended language network, in addition to the sensory-derived representations, leading to a sketch of a dual-coding knowledge neural framework.

Streamgraph Generation with Enhanced Readability

<p indent=0mm>In order to enhance the readability of the streamgraph and assist users identify the target layer more efficiently, a method to optimize the color assignment and label layout in streamgraph based on the existing layout of streamgraph is proposed. First, the color assignment evaluation function of the streamgraph is established according to the layout, and then the color assignment result is got which can maximize the color difference between layers as well as the luminance difference between the background color and layers through genetic algorithm. Secondly, the sliding window is used to traverse each layer, and the tilt angle of each label is automatically adjusted by least squares to fit the shape change of each layer. Then the label layout results are obtained with adaptive label size. The method is compared with existing methods on multiple real and simulated datasets through user study and quantitative evaluation. The results show that the method can significantly improve the recognition accuracy, reduce the recognition time of the layers, and increase the label size at the same time, which means it can improve the readability of the streamgraph effectively.

Color codes with twists: Construction and universal-gate-set implementation

Twists are defects in the lattice that can be used to perform encoded computations. Three basic types of twists can be introduced in color codes, namely, twists that permute color, charge of anyons and domino twists that permute the charge label of an anyon with a color label. In this paper, we study a subset these twists from coding theoretic viewpoint. Specifically, we discuss systematic construction of charge permuting and color permuting twists in color codes. We show that by braiding alone, Clifford gates can be realized in color codes with charge permuting twists. We also discuss implementing single qubit Clifford gates by Pauli frame update and CNOT gate by braiding holes around twists in color codes with color permuting twists. Finally, we also discuss implementing a non-Clifford gate by state injection, thus completing the realization of a universal gate set.

Automated Extraction of Phenotypic Leaf Traits of Individual Intact Herbarium Leaves from Herbarium Specimen Images Using Deep Learning Based Semantic Segmentation.

With the increase in the digitization efforts of herbarium collections worldwide, dataset repositories such as iDigBio and GBIF now have hundreds of thousands of herbarium sheet images ready for exploration. Although this serves as a new source of plant leaves data, herbarium datasets have an inherent challenge to deal with the sheets containing other non-plant objects such as color charts, barcodes, and labels. Even for the plant part itself, a combination of different overlapping, damaged, and intact individual leaves exist together with other plant organs such as stems and fruits, which increases the complexity of leaf trait extraction and analysis. Focusing on segmentation and trait extraction on individual intact herbarium leaves, this study proposes a pipeline consisting of deep learning semantic segmentation model (DeepLabv3+), connected component analysis, and a single-leaf classifier trained on binary images to automate the extraction of an intact individual leaf with phenotypic traits. The proposed method achieved a higher F1-score for both the in-house dataset (96%) and on a publicly available herbarium dataset (93%) compared to object detection-based approaches including Faster R-CNN and YOLOv5. Furthermore, using the proposed approach, the phenotypic measurements extracted from the segmented individual leaves were closer to the ground truth measurements, which suggests the importance of the segmentation process in handling background noise. Compared to the object detection-based approaches, the proposed method showed a promising direction toward an autonomous tool for the extraction of individual leaves together with their trait data directly from herbarium specimen images.

Commercial special bread: Evaluation of texture, color, moisture content, water activity and nutritional labeling

Three types of special bread (WB: bread with whole-wheat flour, LWB: ‘light’ bread with whole-wheat flour, and MB: bread with multi-cereals) were purchased in the local market, and moisture content was analyzed as well as the following parameters water activity (aw), instrumental texture, and instrumental color of the bread (crust and crumb) using spectrophotometer with CIELab system (n=10). The chroma or C * value and the hue angle were also calculated. The labels of the bread were analyzed according to Resolution-RDC nº 360, which provides information about nutritional labeling of packaged foods. The three breads presented in their composition whole wheat flour and wheat fiber, which are indispensable ingredients for the loaves to be considered "special bread". The compression force applied to cause disruption in the bread was higher in WB (3.88 N) without significant difference with MB (3.79 N) and lower in LWB (2.85 N). The color parameters of the breads showed that WB was darker (lower value of L * = 34.44 in the crust) than the others.

Shade recognition of the color label based on the fuzzy clustering

In this article the task of determining the current position of pneumatic actuators is considered. The solution to the given task is achieved by using a technical vision system that allows to apply the fuzzy clustering method to determine in real time the center coordinates and the displacement position of a color label located on the mechatronic complex actuators. The objective of this work is to improve the accuracy of the moving actuator’s of mechatronic complex by improving the accuracy of the color label recognition. The intellectualization of process of the color shade recognition is based on fuzzy clustering. First, a fuzzy model is built, that allows depending on the input parameters of the color intensity for each of the RGB channels and the color tone component, to select a certain color in the image. After that, the color image is binarized and noise is suppressed. The authors used two defuzzification models during simulation a fuzzy system: one is based on the center of gravity method (CoG) and the other is based on the method of area ratio (MAR). The model is implemented based on the method of area ratio and allows to remove the dead zones that are present in the center of gravity model. The method of area ratio determines the location of the color label in the image frame. Subsequently, when the actuator is moved longitudinally, the vision system determines the location of the color label in the new frame. The color label position offset between the source and target images allows to determine the moved distance of the color label. In order to study how noise affects recognition accuracy, the following digital filters were used: median, Gaussian, matrix and binomial. Analysis of the accuracy of these filters showed that the best result was obtained when using a Gaussian filter. The estimation was based on the signal-to-noise coefficient. The mathematical models of fuzzy clustering of color label recognition were simulated in the Matlab/Simulink environment. Experimental studies of technical vision system performance with the proposed fuzzy clustering model were carried out on a pneumatic mechatronic complex that performs processing, moving and storing of details. During the experiments, a color label was placed on the cylinder, after which the cylinder moved along the guides in the longitudinal direction. During the movement, video recording and image recognition were performed. To determine the accuracy of color label recognition, the PSNR and RMSE coefficients were calculated which were equal 38.21 and 3.14, respectively. The accuracy of determining the displacement based on the developed model for recognizing color labels was equal 99.7%. The defuzzifier speed has increased to 590 ns.

Unpredictive linguistic verbal cues accelerate congruent visual targets into awareness in a breaking continuous flash suppression paradigm

One of the most influential ideas within the domain of cognition is that of embodied cognition, in which the experienced world is the result of an interplay between an organism’s physiology, sensorimotor system, and its environment. An aspect of this idea is that linguistic information activates sensory representations automatically. For example, hearing the word ‘red’ would automatically activate sensory representations of this color. But does linguistic information prioritize access to awareness of congruent visual information? Here, we show that linguistic verbal cues accelerate matching visual targets into awareness by using a breaking continuous flash suppression paradigm. In a speeded reaction time task, observers heard spoken color labels (e.g., red) followed by colored targets that were either congruent (red), incongruent (green), or neutral (a neutral noncolor word) with respect to the labels. Importantly, and in contrast to previous studies investigating a similar question, the incidence of congruent trials was not higher than that of incongruent trials. Our results show that RTs were selectively shortened for congruent verbal–visual pairings, and that this shortening occurred over a wide range of cue–target intervals. We suggest that linguistic verbal information preactivates sensory representations, so that hearing the word ‘red’ preactivates (visual) sensory information internally.

Young children’s understanding of ordinal and spatial labels

The goal of this study was to understand the development of ordinal and spatial language during the preschool years. One hundred twenty-nine 3- and 4-year-old children searched for stickers in labeled train cars hidden from view by three identical tunnels. Children were randomly assigned to label condition, comparing ordinal (first, second, third), spatial (front, middle, back), and color labels (red, green, yellow). As expected, search performance improved significantly with age. Importantly, search performance was significantly lower for ordinal labels than for color labels (the control condition), with spatial labels intermediate and not different from the other two conditions. Positive correlations were found between search performance and parent reports of child language. These findings provide important details about developmental improvements in ordinal and spatial language during the preschool years.

Deep Learning Assisted Robotic Magnetic Anchored and Guided Endoscope for Real-Time Instrument Tracking

This letter presents the first case of implementing deep-learning based instrument tracking on a magnetic anchored surgical endoscope. The compact magnetic actuated endoscope has a unique structure that allows operations near the anchor surface, ideal for video assisted thoracoscopic surgery (VATS). Autonomous tool tracking alleviates surgeon's burden and prevents human errors from muscle fatigues or miscommunication. However, conventional methods rely on color labels or require modification to instrument, and has risk of failure due to occlusion of marker. In this letter, we combine deep-learning instrument detection with visual servoing control. This allows the magnetic endoscope to track surgical tools automatically, without color markers or instrument modification. We used a modified TernausNet-16 network that can detect surgical instrument in real time, with a small training dataset of 1846 images. Experiments show that the magnetic endoscope can effectively track a marker-less instrument. It can also track continuous motions of a target traveling at 40 mm $/$ s. The performance was also verified by completing mock-up surgical task in a simulated thoracic cavity.

Will a green color and nature images make consumers pay more for a green product?

Purpose Would a green color label increase the dollar amount consumers are willing to pay for a green product? Would nature images (such as a leaf or flower) on the label have the same effect? This paper aims to examine the role of these labeling strategies in influencing consumer willing to pay. Design/methodology/approach Using a 2 × 3 experiment, the authors empirically test the research questions across two studies: in the controlled-lab setting with 160 students (Study 1) and in a field-setting with 268 consumers shopping at a grocery store (Study 2). Findings Results are consistent across both studies. Surprisingly, participants are willing to pay more for the product when it has a white-toned label rather than a green-toned one. Follow-up path analysis, with Study 2 data, reveals that a white-toned label has both an indirect (through more favorable evaluations of the product’s environmental friendliness), as well as a direct impact on willingness to pay. In providing a post hoc explanation, it is argued that a white-toned label better directs attention towards the claim signaling the product’s eco-friendliness, while providing a “clean”, “high-quality” look. In both studies however, nature images on the label did not have a significant effect. Practical implications Insights are particularly interesting for practitioners seeking to better label/package green products. Originality/value This investigation is the first to empirically examine how color and images on the label influence the dollar amount consumers are willing to pay for a green product. Findings reveal that counter to common belief, the heavy use of the color green on eco-friendly product labels might not be appropriate; a predominantly white-toned label works better.

An audit of patients presenting with clinically benign breast disease to the Helen Joseph Hospital Breast Imaging Unit

ABSTRACT BACKGROUND: Benign breast pathology is a common presenting complaint, and its assessment is important to characterise not to miss malignant pathology. At Helen Joseph Hospital (HJH), patients are triaged at the breast clinic according to the clinical suspicion of benign versus malignant disease. The patients are assigned a colour label based on their clinical presentation. This triage system affects waiting times between clinical examination and mammography appointments. This study aims to assess the association between clinical examination and the radiological and pathological findings of disorders deemed clinically benign, and to ascertain the spectrum of benign breast disorders encountered at HJH METHOD: A retrospective study of imaging results of patients at HJH presenting as clinically benign breast disorders from January to June 2018 was conducted. Assessed Breast Imaging-Reporting and Data System (BI-RADS) score was noted and if core biopsies were performed, their results and patient demographics were documented RESULTS: Of the 1 263 clinically benign patients presenting from January to June 2018, the radiological assessment was: BI-RADS 1: 158 (12.5%), BI-RADS 2: 685 (54.2%), BI-RADS 3: 292 (23.1%), BI-RADS 4a: 54 (4.3%), BI-RADS 4b: 29 (2.3 %), BI-RADS 4c: 21 (1.7%), BI-RADS 5: 24 (1.9%). There were 133 biopsies (including eight BI-RADS 3 patients), with 46 (3.6%) confirmed malignancies. The combined specificity of mammography and ultrasound was 65.52% (54.56-75.39%) and combined sensitivity 91.30% (79.21-97.58% CONCLUSION: There is a vast spectrum of benign conditions presenting in this population group with only 3.6% confirmed malignancies, confirming an accurate triage system utilised at the breast clinic. Radiological imaging is highly sensitive but less specific, emphasising the triad of clinical, radiological and histological assessment as the gold standard with regard to diagnosis of breast disease Keywords: clinically benign breast disease, breast imaging unit, Helen Joseph Hospital

Art Image Processing and Color Objective Evaluation Based on Multicolor Space Convolutional Neural Network.

A convolutional neural network's weight sharing feature can significantly reduce the cumbersome degree of the network structure and reduce the number of weights that need to be trained. The model can directly input the original image, without the process of feature extraction and data reconstruction in common classification algorithms. This kind of network structure has got a good performance in image processing and recognition. Based on the color objective evaluation method of the convolutional neural network, this paper proposes a convolutional neural network model based on multicolor space and builds a convolutional neural network based on VGGNet (Visual Geometry Group Net) in three different color spaces, namely, RGB (Red Green Blue), LAB (Luminosity a b), and HSV (Hue Saturation Value) color spaces. We carry out research on data input processing and model output selection and perform feature extraction and prediction of color images. After a model output selection judger, the prediction results of different color spaces are merged and the final prediction category is output. This article starts with the multidimensional correlation for visual art image processing and color objective evaluation. Considering the relationship between the evolution of artistic painting style and the color of artistic images, this article explores the characteristics of artistic image dimensions. In view of different factors, corresponding knowledge extraction strategies are designed to generate color label distribution, provide supplementary information of art history for input images, and train the model on a multitask learning framework. In this paper, experiments on multiple art painting data sets prove that this method is superior to single-color label classification methods.

Skin color and race.

Skin color is the primary physical criterion by which people have been classified into groups in the Western scientific tradition. From the earliest classifications of Linnaeus, skin color labels were not neutral descriptors, but connoted meanings that influenced the perceptions of described groups. In this article, the history of the use of skin color is reviewed to show how the imprint of history in connection with a single trait influenced subsequent thinking about human diversity. Skin color was the keystone trait to which other physical, behavioral, and culture characteristics were linked. To most naturalists and philosophers of the European Enlightenment, skin color was influenced by the external environment and expressed an inner state of being. It was both the effect and the cause. Early investigations of skin color and human diversity focused on understanding the central polarity between “white” Europeans and nonwhite others, with most attention devoted to explaining the origin and meaning of the blackness of Africans. Consistently negative associations with black and darkness influenced philosophers David Hume and Immanuel Kant to consider Africans as less than fully human and lacking in personal agency. Hume and Kant's views on skin color, the integrity of separate races, and the lower status of Africans provided support to diverse political, economic, and religious constituencies in Europe and the Americas interested in maintaining the transatlantic slave trade and upholding chattel slavery. The mental constructs and stereotypes of color‐based races remained, more strongly in some places than others, after the abolition of the slave trade and of slavery. The concept of color‐based hierarchies of people arranged from the superior light‐colored people to inferior dark‐colored ones hardened during the late seventeenth century and have been reinforced by diverse forces ever since. These ideas manifest themselves as racism, colorism, and in the development of implicit bias. Current knowledge of the evolution of skin color and of the historical development of color‐based race concepts should inform all levels of formal and informal education. Awareness of the influence of color memes and race ideation in general on human behavior and the conduct of science is important.

Showing pesticides’ true colors: The effects of a farmer-to-farmer training program on pesticide knowledge

Information plays an important role in technology adoption and behavior change. This may be particularly true for pesticide decisions, which affect farmer income, health, and the environment. Pesticide use is increasing, and farmers often have incomplete knowledge of how pesticides affect crop production and health. Researchers regularly recommend information interventions to improve pesticide knowledge. This study uses a randomized controlled trial of a pesticide training program in Zambia to identify the effects of information on farmer knowledge of both the health risks and the production benefits of pesticides. We use detailed knowledge assessments covering identification and interpretation of toxicity color labels, pest control properties, and exposure to reveal a more complete picture of pesticide knowledge and to identify where information can have the greatest impact. Our results show that training significantly improved farmer knowledge, particularly for the more complicated pesticide characteristics that are harder to learn from experience. We also find that more experienced farmers had worse prior knowledge of pesticide toxicity, though they also had larger increases from training. Overall, our results suggest that pesticide information interventions should focus on the pest control properties of pesticides and on showing pesticides’ true colors through toxicity identification by color labels.