Fixed typo

README.md

@@ -1,37 +1,48 @@
-# Unified Perception: Efficient Video Panoptic Segmentation with Minimal Annotation Costs
+# UniPercept
 
-Welcome to the PyTorch 2 implementation of Unified Perception, an innovative approach to depth-aware video panoptic segmentation. Our method achieves state-of-the-art performance without the need for video-based training. Instead, it utilizes a two-stage cascaded tracking algorithm that reuses object embeddings computed in an image-based network. This repository mirrors the research paper [Unified Perception](https://arxiv.org/abs/2303.01991).
+## Installation
 
-## Introduction
+This package requires at least Python 3.11 and PyTorch 2.1. Once you have created an environment with these 
+dependencies, we can proceed to install `unipercept` using one of three installation methods.
 
-Unified Perception is designed to tackle the challenge of depth-aware video panoptic segmentation with high precision. The method proposed in this repository effectively combines the benefits of image-based networks with a two-stage cascaded tracking algorithm, ultimately enhancing the performance of video panoptic segmentation tasks without the need for video-based training.
-
-## Technical Documentation
-
-For a comprehensive understanding of the Unified Perception implementation, please visit our technical documentation hosted on:
-
-[https://tue-mps.github.io/unipercept](https://tue-mps.github.io/unipercept)
-
-## Usage & Installation (Coming Soon)
+### Stable release (recommended)
+You can install the latest stable release from PyPI via
+```bash
+pip install unipercept
+```
 
-_This section is under development and will be added in the near future. It will provide step-by-step instructions for using and installing the Unified Perception package. Consider starring or subscribing to this repository for updates._
+### Master branch
+To install the latest version, which is not guaranteed to be stable, install from GitHub using 
+```bash
+pip install git+https://github.com/kurt-stolle/unipercept.git
+```
 
-The following _optional_ packages may be installed:
+### Developers
+If your use-case requires changes to our codebase, we recommend that you first fork this repository and download your
+own fork locally. Assuming you have the GitHub CLI installed, you can clone your fork with
+```bash
+gh repo clone unipercept
+```
+Then, you can proceed to install the package in editable mode by running
 ```bash
-pip uninstall pillow
-CC="cc -mavx2" pip install -U --force-reinstall pillow-simd
+pip install --editable unipercept
 ```
+You are invited to share your improvements to the codebase through a pull request on this repository. 
+Before making a pull request, please ensure your changes follow our code guidelines by running `pre-commit` before 
+adding your files into a Git commit.
 
 ## Training and evaluation
 
-Use the CLI:
+Models can be trained and evalurated from the CLI or through the Python API. 
+
+### CLI
+To train a model with the CLI:
 ```bash
-unicli train --config <config name>
+unicli train --config <config path>
 ```
+Without a `<config name>`, an interactive prompt will be started to assist in finding a configuration file.
 
-Without a `<config name>`, an interactive prompt will be started.
-
-## Validation and testing
+## Developer guidelines
 All tests can ran via `python -m pytest`. 
 However, we also provide a `make` directive that uses `pytorch-xdist` to speed up the process:
 ```

sources/unipercept/integrations/wandb_integration.py

@@ -101,9 +101,9 @@ def on_trackers_setup(self, params: EngineParams, state: State, control: Signal,
     def on_save(
         self, params: EngineParams, state: State, control: Signal, *, model_path: str, state_path: str, **kwargs
     ):
-        if self.model_history <= 0:
+        if self.model_history > 0:
             self._log_model(model_path)
-        if self.state_history <= 0:
+        if self.state_history > 0:
             self._log_state(state_path)
 
     @TX.override
@@ -118,7 +118,7 @@ def on_inference_end(
         results_path: str,
         **kwargs,
     ):
-        if self.inference_history <= 0:
+        if self.inference_history > 0:
             self._log_inference(results_path)
         if self.tabulate_inference_timings:
             self._log_profiling("inference/timings", timings)

sources/unipercept/nn/backbones/timm.py

@@ -60,6 +60,7 @@ def __init__(
         pretrained: bool = True,
         nodes: T.Sequence[int] | None | int = None,
         keys: T.Sequence[str] | None = None,
+        jit_script: bool = False,
         **kwargs,
     ):
         dims = _get_dimension_order(name)
@@ -78,7 +79,10 @@ def __init__(
 
         super().__init__(dimension_order=dims, feature_info={k: v for k, v in zip(keys, info)}, **kwargs)
 
-        self.ext = torch.jit.script(extractor)
+        self.ext = extractor
+
+        if jit_script:
+            self.ext = torch.jit.script(self.ext)  # type: ignore
 
     @override
     def forward_extract(self, images: torch.Tensor) -> OrderedDict[str, torch.Tensor]: