Hazard recognition
Methods, systems, and devices are provided for identifying hazards. According to one aspect, a computer-implemented method can include receiving a plurality of sensor data including one or more image files from a mobile device. The method can include generating one or more position and label pairs based on the plurality of sensor data. The method can include assigning a hazard recognition to each of the position and label pairs. The method can include assigning a score associated to each of the hazard recognitions. The method can include displaying a result including one or more image results based on the one or more image files, one or more hazard recognitions, the one or more hazard recognitions associated with at least one of the one or more image results, and one or more scores associated to each of the hazard recognitions.
1. A computer-implemented method of identifying hazards, the method comprising:
receiving a plurality of sensor data including one or more image files from a mobile device;
generating one or more position and label pairs based on the plurality of sensor data;
training an artificial intelligence (AI) model based on a training data set, wherein the training data set comprises a plurality of images of hazards labeled according to a plurality of hazard types;
generating, via the trained AI model, a hazard recognition to be assigned to each of the position and label pairs;
generating, via the trained AI model, a score to be associated with each of the hazard recognitions; and
displaying a result comprising:
one or more image results based on the one or more image files;
one or more hazard recognitions, the one or more hazard recognitions associated with at least one of the one or more image results and a plurality of different hazard types; and
one or more scores associated to each of the different hazard types of the hazard recognitions.
2. The method of claim 1 , wherein the sensor data is comprised of video capture data from a video or imaging capture device, imaging angle from an accelerometer of the mobile device, user selected data, or a combination thereof.
3. The method of claim 2 , wherein the user selected data are derived from selectable themes or categories on a mobile application.
4. The method of claim 3 , wherein the mobile application comprises a user interface configured to receive selections by the user and give recommendations, or recommended selections, or a combination thereof.
5. The method of claim 4 , wherein the recommendations are based on a detection of a minimum or maximum threshold based on a camera angle of the mobile device when the user is using the mobile application.
6. The method of claim 1 , wherein the label identifies a characteristic of an item, scene, path, condition, potential hazards, or a combination thereof.
7. The method of claim 1 , wherein the position identifies a local position of an item, scene, path, condition, potential hazards, or a combination thereof.
8. The method of claim 7 , wherein the local position is identified based on an entire frame of an image generated by the sensor data.
9. The method of claim 7 , wherein the local position is identified based on a portion of an image associated to a bounding box or a segmented list of points within the image.
10. The method of claim 1 , further comprising:
applying a recognition filter to each pair of the one or more position and label pairs; and
generating one or more filtered label and position pairs.
11. The method of claim 10 , wherein applying the recognition filter to each pair of the one or more position and label pairs comprises applying a hysteresis filtering.
12. The method of claim 10 , wherein applying the recognition filter to each pair of the one or more position and label pairs comprises context recognition filtering of a user based on a selected context provided by the user.
13. The method of claim 10 , further comprising:
generating a virtual hazard recognition based on the generating of the one or more filtered label and position pairs.
14. The method of claim 1 , wherein training the AI model comprises:
receiving a plurality of video files from a crowdsourcing platform;
processing each video file into a set of image files;
sending each image file of the set of image files into an image queue; processing the image queue comprising labelling each image file according to a hazard characterization each image file represents; and
tagging each of the image files with one or more of the hazard characterizations.
15. The method of claim 1 , wherein the hazard recognition is assigned based on a user's personal data including age, health, vision, physical mobility, personal medical information and conditions, or a combination thereof.
16. The method of claim 1 , further comprising:
associating one or more hazard recognitions and score of the hazard recognitions with a scene or space.
17. The method of claim 16 , further comprising:
generating a hazard score based on an aggregate of each score of the hazard recognitions associated with the scene or space.
18. The method of claim 1 , further comprising:
generating one or more corrective measures based on at least part of the one or more scores associated with the hazard recognitions, and
displaying the one or more corrective measures on the mobile device.
19. The method of claim 1 , further comprising:
receiving a selection input from a human agent to determine whether a hazard classification based on one or more of the assigned hazard recognitions has been addressed.