An example apparatus for mining multi-scale hard examples includes a convolutional neural network to receive a mini-batch of sample candidates and generate basic feature maps. The apparatus also includes a feature extractor and combiner to generate concatenated feature maps based on the basic feature maps and extract the concatenated feature maps for each of a plurality of received candidate boxes. The apparatus further includes a sample scorer and miner to score the candidate samples with multi-task loss scores and select candidate samples with multi-task loss scores exceeding a threshold score.

Presented herein are systems and methods that provide for automated analysis of three-dimensional (3D) medical images of a subject in order to automatically identify specific 3D volumes within the 3D images that correspond to specific anatomical regions (e.g., organs and/or tissue). Notably, the image analysis approaches described herein are not limited to a single particular organ or portion of the body. Instead, they are robust and widely applicable, providing for consistent, efficient, and accurate detection of anatomical regions, including soft tissue organs, in the entire body. In certain embodiments, the accurate identification of one or more such volumes is used to automatically determine quantitative metrics that represent uptake of radiopharmaceuticals in particular organs and/or tissue regions. These uptake metrics can be used to assess disease state in a subject, determine a prognosis for a subject, and/or determine efficacy of a treatment modality.

A target object detection model is provided. The target object detection model includes a YOLOv3-Tiny model. Through the target object detection model, low-level information in the YOLOv3-Tiny sub-model can be merged with high-level information therein, so as to fuse the low-level information and the high-level information. Since the low-level information can be further used, the comprehensiveness of target detection is effectively improved, and the detection effect of small targets is improved.

A farming machine includes one or more image sensors for capturing an image as the farming machine moves through the field. A control system accesses an image captured by the one or more sensors and identifies a distance value associated with each pixel of the image. The distance value corresponds to a distance between a point and an object that the pixel represents. The control system classifies pixels in the image as crop, plant, ground, etc. based on depth information in in the pixels. The control system generates a labelled point cloud using the labels and depth information, and identifies features about the crops, plants, ground, etc. in the point cloud. The control system generates treatment actions based on any of the depth information, visual information, point cloud, and feature values. The control system actuates a treatment mechanism based on the classified pixels.

Selection of on optical pattern in a scene is identified by overlaying, on a display, an indicator of a detected optical pattern identifying a location of the optical pattern in one or more images, receiving a user input on the display at a position that does not overlap the location of the optical pattern, and presenting information related to the optical pattern, based on receiving the user input, even though the position of user input did not overlap the location of the optical pattern. The user input can be received at a detached selection indicator and/or using an adaptive input area.

Presented herein are systems and methods that provide for automated analysis of three-dimensional (3D) medical images of a subject in order to automatically identify specific 3D volumes within the 3D images that correspond to specific anatomical regions (e.g., organs and/or tissue). Notably, the image analysis approaches described herein are not limited to a single particular organ or portion of the body. Instead, they are robust and widely applicable, providing for consistent, efficient, and accurate detection of anatomical regions, including soft tissue organs, in the entire body. In certain embodiments, the accurate identification of one or more such volumes is used to automatically determine quantitative metrics that represent uptake of radiopharmaceuticals in particular organs and/or tissue regions. These uptake metrics can be used to assess disease state in a subject, determine a prognosis for a subject, and/or determine efficacy of a treatment modality.

The techniques described herein relate to methods, apparatus, and computer readable media configured to test a pose of a three-dimensional model. A three-dimensional model is stored, the three dimensional model comprising a set of probes. Three-dimensional data of an object is received, the three-dimensional data comprising a set of data entries. The three-dimensional data is converted into a set of fields, comprising generating a first field comprising a first set of values, where each value of the first set of values is indicative of a first characteristic of an associated one or more data entries from the set of data entries, and generating a second field comprising a second set of values, where each second value of the second set of values is indicative of a second characteristic of an associated one or more data entries from the set of data entries, wherein the second characteristic is different than the first characteristic. A pose of the three-dimensional model is tested with the set of fields, comprising testing the set of probes to the set of fields, to determine a score for the pose.

A computer-implemented method for generating a rotation-invariant feature descriptor for a location in an image for use in performing descriptor matching in analysing the image, extracts samples according to a descriptor pattern for the location in the image; uses the extracted samples to determine a measure of rotation for the location in the image, the measure of rotation describing an angle between an orientation of the image and a characteristic direction of the image at the location; generating a feature descriptor for the location in the image by determining a set of samples characterising the location in dependence on the determined measure of rotation and the extracted samples; and processes the determined set of samples to generate the feature descriptor for the location in the image.

A farming machine includes one or more image sensors for capturing an image as the farming machine moves through the field. A control system accesses an image captured by the one or more sensors and identifies a distance value associated with each pixel of the image. The distance value corresponds to a distance between a point and an object that the pixel represents. The control system classifies pixels in the image as crop, plant, ground, etc. based on depth information in in the pixels. The control system generates a labelled point cloud using the labels and depth information, and identifies features about the crops, plants, ground, etc. in the point cloud. The control system generates treatment actions based on any of the depth information, visual information, point cloud, and feature values. The control system actuates a treatment mechanism based on the classified pixels.

An image text recognition method includes converting an image into a grayscale image, and segmenting, according to layer intervals to which grayscale values of pixels in the grayscale image belong, the grayscale image into grayscale layers with one corresponding to a layer interval, performing image erosion on a grayscale layer to obtain a feature layer corresponding to the grayscale layer, the feature layer including at least one connected region; overlaying feature layers to obtain an overlaid feature layer, the overlaid feature layer including connected regions; dilating connected regions on the overlaid feature layer according to a preset direction to obtain text regions; and performing text recognition on the text regions on the overlaid feature layer to obtain a recognized text corresponding to the image.