I set up a VTON model in Gradio that takes 2 images as input, and returns an image of the person wearing the clothes submitted in the previous 2 images. When running the API recorder within Gradio, it gives me a script which will mimic the call just made: from gradio_client import Client, file client = Client("https://5c47fedb2c3a49c2e8.gradio.live/") client.predict( dict={"background":handle_file('https://5c47fedb2c3a49c2e8.gradio.live/gradio_api/file=/tmp/gradio/b85624d24ac163fd1e2f6e603372d24ea223825f1eb1a72a60a9d30d44b39d85/background.png'),"layers":[handle_file('https://5c47fedb2c3a49c2e8.gradio.live/gradio_api/file=/tmp/gradio/3b5a9ef2815439980a0f8b5725869b157a4bea163aa3c9a628614c26bfac1141/layer_0.png')],"composite":handle_file('https://5c47fedb2c3a49c2e8.gradio.live/gradio_api/file=/tmp/gradio/b85624d24ac163fd1e2f6e603372d24ea223825f1eb1a72a60a9d30d44b39d85/composite.png')}, garm_img=handle_file('https://5c47fedb2c3a49c2e8.gradio.live/gradio_api/file=/tmp/gradio/db8f2f2d169bf74efb38a963ef0b8cc2a6f23f4ebe574026df737ec5ca10c3d0/guccimane.jpg'), garment_des="", category="upper_body", is_checked=True, is_checked_crop=True, denoise_steps=20, is_randomize_seed=True, seed=1, number_of_images=1, api_name="/tryon" ) This apparently hits XXX.gradio.live/tryon and is able to perform inference on the two images. The API reference also denotes /tryon as an endpoint, seemingly for the exact use case I want to use it for (external API usage). Unfortunately, every response I get from any external source (cURL in bash, Postman, Python from my desktop) is just a 404 error. For some reason, it cannot see this endpoint, and I have no idea why. Any suggestions as to why this endpoint isn't exposed? Am I just hitting the wrong URL, and if so how could I find the correct one? Continue reading...
