Image processing systems can include one or more cameras configured to obtain image data, one or more memory devices configured to store a classification model that classifies image features within the image data as including or not including detected objects, and a field programmable gate array (FP
Image processing systems can include one or more cameras configured to obtain image data, one or more memory devices configured to store a classification model that classifies image features within the image data as including or not including detected objects, and a field programmable gate array (FPGA) device coupled to the one or more cameras. The FPGA device is configured to implement one or more image processing pipelines for image transformation and object detection. The one or more image processing pipelines can generate a multi-scale image pyramid of multiple image samples having different scaling factors, identify and aggregate features within one or more of the multiple image samples having different scaling factors, access the classification model, provide the features as input to the classification model, and receive an output indicative of objects detected within the image data.
대표청구항▼
1. An image processing system, comprising: one or more cameras configured to obtain image data;one or more memory devices configured to store a classification model that classifies image features within the image data as including or not including detected objects; anda field programmable gate array
1. An image processing system, comprising: one or more cameras configured to obtain image data;one or more memory devices configured to store a classification model that classifies image features within the image data as including or not including detected objects; anda field programmable gate array (FPGA) device coupled to the one or more cameras, the FPGA device configured to implement one or more image processing pipelines for image transformation and object detection;the one or more image processing pipelines including a plurality of logic blocks and interconnectors programmed to: generate a multi-scale image pyramid of multiple image samples having different scaling factors; identify and aggregate features within one or more of the multiple image samples having different scaling factors; access the classification model stored in the one or more memory devices; provide the features within the one or more of the multiple image samples as input to the classification model; and produce an output indicative of objects detected within the image data;wherein the features identified and aggregated within the one or more of the multiple image samples comprise edge portions, and wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors that are programmed to determine an angle classification for each of the identified edge portions, and to assign the edge portion to one of a plurality of different bins depending on the angle classification for that edge portion. 2. The image processing system of claim 1, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors that are programmed to determine a histogram descriptive of the plurality of different bins. 3. The image processing system of claim 2, wherein the histogram comprises a histogram of oriented gradients for the identified edge portions. 4. The image processing system of claim 1, wherein the plurality of different bins are defined to have different sizes based on an amount of image data in each image sample such that bin sizes are smaller for image samples having a greater amount of image data. 5. The image processing system of claim 1, wherein the one or more image processing pipelines include a plurality of logic blocks and interconnectors programmed to generate one or more channel images from the image data, each channel image corresponding to a feature map that maps a patch of one or more input pixels from the image data to an output pixel within the channel image. 6. The image processing system of claim 1, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors designed to convert intermediate stages of the image data from the one or more cameras from a floating point representation to fixed point integer-based representation. 7. The image processing system of claim 1, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors programmed to convert the image data from the one or more cameras into a multi-parameter representation including values corresponding to an image hue parameter, an image saturation parameter, and an image greyscale parameter. 8. The image processing system of claim 1, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors programmed to convert the image data from a representation having multiple color components to a greyscale representation. 9. The image processing system of claim 1, wherein the one or more image processing pipelines include a plurality of logic blocks and interconnectors programmed to determine a sliding window of fixed size, to analyze successive image patches within each of the multiple image samples using the sliding window of fixed size, and to identify objects of interest within the successive image patches. 10. A vehicle control system, comprising: one or more cameras configured to obtain image data within an environment proximate to a vehicle;a field programmable gate array (FPGA) device coupled to one or more cameras, the FPGA device configured to implement one or more image processing pipelines for image transformation and object detection, the one or more image processing pipelines including a plurality of logic blocks and interconnectors programmed to: generate from the image data a multi-scale image pyramid of multiple image samples having different scaling factors; identify and aggregate features within one or more of the multiple image samples having different scaling factors; and to detect one or more objects of interest within the multiple image samples based at least in part on the features, wherein the features identified and aggregated within the one or more of the multiple image samples comprise edge portions, and wherein the second image processing pipeline for object detection further includes a plurality of logic blocks and interconnectors that are programmed to determine an angle classification for each of the identified edge portions, and to assign the edge portion to one of a plurality of different bins depending on the angle classification for that edge portion;one or more computing devices configured to receive an output from the FPGA device and to further characterize the objects of interest. 11. The vehicle control system of claim 10, wherein the one or more computing devices are further configured to control motion of the vehicle based at least in part on the one or more objects of interest detected within the image data from the one or more cameras. 12. The vehicle control system of claim 10, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors that are programmed to determine a histogram of oriented gradients descriptive of the plurality of different bins. 13. The vehicle control system of claim 10, wherein the plurality of different bins are defined to have different sizes based on an amount of image data in each image sample such that bin sizes are smaller for image samples having a greater amount of image data. 14. The vehicle control system of claim 10, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors programmed to convert the image data from the one or more cameras into a multi-parameter representation including values corresponding to an image hue parameter, an image saturation parameter, and an image greyscale parameter. 15. The vehicle control system of claim 10, wherein the one or more image processing pipelines further include a plurality of logic blocks and interconnectors programmed to determine a sliding window of fixed size, to analyze successive image patches within each of the multiple image samples using the sliding window of fixed size, and to identify objects of interest within the successive image patches.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (10)
Stein, Gideon P.; Ferencz, Andras D.; Avni, Ofer, Estimating distance to an object using a sequence of images recorded by a monocular camera.
Sanjay Devappa Rai ; Nicholas Barton ; Troy Taylor ; Xueming Henry Gu, Primary and secondary color manipulations using hue, saturation, luminance and area isolation.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.