Train me AI based model (on CNN) with great accuracy
Betalt ved levering
We are seeking an experienced AI model developer to create a high-precision and fast traffic monitoring and safety AI CNN model. We prioritize accuracy and are interested in leveraging public datasets specific to Pakistan or India to ensure relevance, as we do not have access to the dataset.
The model's primary function will be to analyze images from traffic signals and provide valuable insights and alerts. The model should be capable of the following tasks and seamlessly integrate with a Node.js backend:
Model should be CNN.
1) Accurately detect and count various types of vehicles, including cars, bikes, rickshaws, and any other vehicle types present in the image.
2) Provide the total vehicle count on the road.
3) Calculate the percentage of empty road and the percentage of road with vehicles accurately, potentially using comparison techniques.
Ambulance and Police Car Detection:
1) Identify and count the number of ambulances and police cars in the image, with a priority on safety.
2) Determine if an accident has occurred within the image and return "isAccidentOccured = true" if detected; otherwise, return "false."
Suspicious Object Recognition:
3) Recognize and flag the presence of suspicious objects, such as weapons (e.g., guns), within the image.
Return "isSuspesiousObject = true" if a suspicious object is detected; otherwise, return "false."
Projekt ID: #37255609