Fall detection using camera
Budget: ₹400 – ₹750 INR
My ESP32-CAM streams live frames over Wi-Fi to a PC where a Python demo—sourced entirely from an open-source project—runs basic fall-detection. I now need noticeably higher accuracy and lower false-positives, ideally through a Convolutional Neural Network.
What I need you to do
• Audit the existing Python code, dataset handling, and real-time feed integration.
• Decide whether fine-tuning the current CNN or building a new one from scratch (TensorFlow, PyTorch, or another Python-friendly framework) will give the best accuracy-to-latency balance.
• Train, validate, and test the model; then wire it back into the live ESP32 stream so detection happens on the PC in real time.
• Supply the trained weights, the updated script, and concise setup notes so I can reproduce the results on my machine.
Acceptance will be based on demonstrable accuracy improvement against my current test clips and smooth real-time performance when the camera is live.
I am looking for only Gujarat, India based developer only so please any other person won't be prefer.
What I need you to do
• Audit the existing Python code, dataset handling, and real-time feed integration.
• Decide whether fine-tuning the current CNN or building a new one from scratch (TensorFlow, PyTorch, or another Python-friendly framework) will give the best accuracy-to-latency balance.
• Train, validate, and test the model; then wire it back into the live ESP32 stream so detection happens on the PC in real time.
• Supply the trained weights, the updated script, and concise setup notes so I can reproduce the results on my machine.
Acceptance will be based on demonstrable accuracy improvement against my current test clips and smooth real-time performance when the camera is live.
I am looking for only Gujarat, India based developer only so please any other person won't be prefer.