Low level, multi threaded, http API service for image recognition. Written in Rust with CUDA support.
This software requires:
-
Rust nightly: rustc 1.40.0-nightly, cargo 1.40.0-nightly, rustup 1.20.2
-
and more (please see Cargo.toml)
-
please if building for IoT use proper compilation settings or build on device.
-
please compile with
IS_CUDA = false
first, or check if your hardware is capable of CUDA before compiling
-
Download weights
$ wget -c https://github.com/LaurentMazare/ocaml-torch/releases/download/v0.1-unstable/yolo-v3.ot
toassets
folder. or runassets/downloads.sh
-
use nightly:
rustup default nightly
- cargo:
$ RUST_LOG=info ROCKET_ENV=development cargo run
- docker:
docker-compose -f docker-compose.dev.yml up
- cargo:
$ RUST_LOG=error ROCKET_ENV=production cargo run
- docker:
docker-compose up
$ cargo build --release
thenchmod +x ./target/release/hebbian_mirror
then$ RUST_LOG=error ROCKET_ENV=production ./target/release/hebbian_mirror
Python script:
- python3 -m venv venv
, then source venv/bin/activate
,
then pip install -r requirements.txt
and run test python e2e_test.py
- Lenart Bartosz
- Claire Amalfitano
- Janusz Roll
Very good explanation from Ayoosh Kathuria